Yes. Otherwise each side will sacrifice a unit of 5 infantry. Hell, one side could even involve a militia unit.
How do you deal with the fact that total CS changes over time? Sometimes radically, such as after a huge battle. Does this include militia, or only mobile? If it includes mobile only, then small realms under siege from larger ones are screwed, because they rely on large amounts of militia. If it includes militia, then just about everyone is screwed, as nearly every realm has significant militia forces somewhere.
But to get the full bonus (or avoid the penalty) in your proposal, you have to have your entire army constantly involved in battle. We already get enough complaints that war isn't exciting enough, and nothing but a continuous cycle of refit/smash/repeat. Doing something like this enforces, or even mandates, this undesirable cycle.
Any way you slice it, this type of proposed mechanic will really slow things down, and is sure to screw over someone who's not fighting the type of war you are spec'ing out with your proposal.
I might have missed something in the initial proposal, but who cares how attainable the "full bonus" is? Even if you could have gotten more, you'd still have gotten one, thus not much to complain about.
If the bonuses stack (as long as they cap), two battles would mean the same as one of twice the size. No way to abuse this. Also, war in BM IS a continuous cycle of refit/smash/repeat. What else do you expect in a system where you can only recruit in your capital (and repair/cash in bonds in your realm's cities, most likely to be your capital) and where moral penalties are inflicted for being away from your lands too long? It always was such a cycle, and will remain as long as recruiting in capital restrictions remain. Bonuses due to conflict don't change anything in this regards.
The rest of the issues you bring up are trivial design issues, not balance issues. The result is pretty much the same in the end, as long as you select the base factor and cap accordingly.
Let's take your example: What would happen if a small realm is under siege and has a lot of militia, if only mobile troops are counted? Let's say they have 7500 CS of militia, and 2500 CS of mobile forces in their capital, and get attacked: 100% of their mobile forces are engaged, thus, they'd get the full 1X bonus to recruitment. If militia is counted? 100% of their troops are defending the capital, so full 1X bonus again. Being a small realm with their capital under siege, they are unlikely to have militia and troops elsewhere.
Now, big realm gets attacked in a region where it has 5000CS mobile and 2500CS militia, out of a total realm-wide of 10000CS mobile and 12500CS militia. Mobile only: the realm gets .5X bonus. With militia counted, they get .25X instead. Without considering the "echo" battles on the next rounds. The big realms get screwed over, you say, because it is harder to get the full bonus? Who cares for the full bonus? They are big, even if the bonus multiplier is small, the effect is bigger than a bigger multiplier in a small realm.
And how would more recruitment slow things down?
Now, I can't say I'm convinced this mechanic would really bring anything worthwhile to the game, but I really have a hard time seeing it as overly complicated or game-breaking either.
