I can mod in these suggestions into the game, so you could see it in the game if you would like. But it would definitely require a lot more quests to be made...

What I'm going to write into the game is randomly infinite spawning monster lairs... just figuring the appropriate mechanics for this and make it difficult enough, but not too easy either.

This can be done as well in the current system of the game if you so desire. Given the current system you can set it up so that when you build an "inn" in your city it will spawn quests every so often for you to investigate via a rumormill type situation. But again this would require people to make a lot more quests because 80 +35 is just not enough to handle this load of quests.

ok. i see your point. you think the tree has to be filled. i disagree with that. i tend to view the trees as open end. a lazy implementation would have been to just have one tree for each path and everyone can reach the end. the way the trees are shaped, there are multiple (many, actually) ways to move through the trees. aim for the high tier traits, pick up several low tier items, combine path traits with magic spellbooks. all of those options can be viable, depending on what you actually want to use this hero for. 

in my opinion, you can't measure "low" level or "mid" level by comparing the levels to the amount of traits in the tree. this may work for other games with more linear trees, but it's quite obvious to me that this isn't the case for this game. i like most of the trees in FH, because they have several viable routes each. you won't fill out the whole tree in one playthrough, you can try a different route next time. i think replayablilty is a big strength of that game, and the tree structure adds to this.

the other thing is that "legendary" is subjective. my champions sure feel legendary to me when they root out that ignis lair 5 tiles away from my capital long before i could have done this with trained units alone, or when they slay a dragon that's going to raze one of my cities. or when they conquer a wildland. it's a matter of perspective, i guess. not really something you can discuss.

i don't think i actually have to explain the properties of dragons. just look them up if you don't know. in short - no, trained units can't deal with dragons. not at the stage of the game when your heroes are level 7, at least. later probably, but that's totally beside the point. can't tell that dragon to wait 100 turns so i can later roflstomp it with companies of mounted pikemen. why would i even want to do that? part of the fun (to me, anyway) is doing the encounters while they are challenging. i'm aware that it's often not necessary. you can, for the most part, just turtle and build up until your troops are strong enough to clear the monsters. but where's the fun in that? you want legendary heroes? use them and do legendary stuff. 

interesting questions (regarding the troops). i think they didn't put that much thought into the XP system for troops. probably because it has a lot less impact then XP for heroes. troops don't compete with heroes due to the levels they may have gained. a level 10 spearman unit is still weak, even though it may now have ~50 HP instead of the initial ~25 and 70 accuracy instead of the original 60. it's the tech upgrades that make the troops powerful, their level just makes them a bit tougher (which is to some extent negated by tech advancement, considering that later on you also unlock a few items that add extra HP to new units and buildings that make them start at a higher level)

@Azunai_
It is funny the way we look at the game and how different our view points are. When I attack an AI city and one shot kill his heroes with my troops. They don't seem all that heroic. A level 7 Hero has only had 5 levels of development (pick class at 2) and the first three or four levels of that is selecting bonus XP traits, though I'm guessing you don't do that. To me, they are just getting started, but you see them as godly because they can help clear an ignis lair, even though you could have done the same thing with standard troops easier. Note that I'm not saying my way is right and yours wrong, I'm just pointing out the differences. You see the scaled up difficulty as fun, where as I see the scaled up difficulty as cumbersome because the difficulty was a by product of unintended consequences. Life is truly funny sometimes.

No, I think that the branch of the tree your champion is specialized in should be able to be mostly filled at mid-level, and after that the champion doesn't really need to progress further, although it could occasionally be nice to have a very powerful champion with a more or less completely filled trait tree even if it comes at the expense of the development of other champions. Instead, we have the ability to mostly fill the specialization branch of one of our champions at the expense of the development of our other champions, or just begin specializing multiple champions.

It's fine if it takes a lot of effort to completely fill the trait tree, but champions more or less stop developing around level 10 or 12, depending on how many you are trying to develop and how much experience is available, and at that point they've barely touched their specializations. Making things worse, the specialized branches of the trait trees tend to be deep and narrow, meaning that all the champions who specialize in something similar end up having mostly identical traits, because there are either no branches to take, or the branches are very short, or the branches only start high up in the tree. It's even worse for your champions if you take anything from the general tree, because the Potential traits don't cover themselves until you'd normally be level 16 or higher and the other traits, while useful for your champion, don't really add anything special or particularly useful (not including developing magic abilities).

At level 10, if you've completely focused a champion on developing one branch of their class tree, you have filled about half to two-thirds of that branch. This makes it feel like the sort of trade-off a low-level character makes - you can drop all of your generalist abilities to get halfway through a specialization, or you can have your general abilities and barely scratch the surface of the specialization - rather than the sort of trade-off a mid-level character makes, where you can drop all of your generalist abilities to more or less completely filling out a specialization, or you can have generalist abilities and get about half-way through your specialization.

How much more linear are the trait trees in those games you're playing? Each of the classes has two, or sometimes three, specialization branches, which either barely branch until you've invested half the levels you're going to have into the specialization, or which have a handful of stubs that don't go anywhere. For that matter, to some degree I'd prefer a lot of short (say, five or six traits), very linear branches to what we currently have - branches 8 traits deep with four or five additional traits strung off of them in random spots, which mostly require the same set of traits to be taken to get to in the first place - because at least then I could start to mix and match pieces of them. Right now, unless I take Tactician at sovereign creation, all my field commanders get Leadership I-III before they can really start to differentiate (and Trainer isn't really that appealing of an alternative to moving through Leadership IV to Tactician, anyways). Governor commanders have a nice variety of choices available, but I don't really want a legendary governor - I want a legendary hero. Damage and curse mages can mix up traits reasonably well, but summoner mages are more or less stuck with one track and the choice between skeletons or high-end summons. Assassins look like they have a nicely branched tree, but dodge bonuses are only really useful early on, and unless I get a good weapon in loot (or pay the ridiculous sums of gold for something from the shop) critical hits are only really keeping the assassin at best on par with the damage dealt by regular troops, and the assassin is generally more fragile than regular troops anyways, and while Executioner might be nice, champions aren't hard enough to kill to make it very useful. Warriors have two very linear lines of development and a bunch of weapon specializations, but weapon specializations are very loot-dependent in utility, the utility of the counterattack bonus is dependent on what you're facing, and the bonus for Mighty Blow is dependent on being able to make use of that ability without over-exposing your warrior to the opposing army, which can depend greatly on the random initial positioning you're stuck with at the start of a battle. And as for Defenders, magic isn't generally a significant enough threat from the AI and monsters to make the spell resistance branch worthwhile (and the spell resistance line also doesn't grant access to plate armor), while the Healer branch is tiny and the Defender branch has only tiny side-branches (which are often only marginally useful).