Crusty

I think it could lead to some interesting situations. Like what if you're sneaking around the castle of a known enemy. You encounter a bunch of guards, you see no alternative and you fight. Their morale breaks as the battle progresses and the last survivor(s) tries to flee, hoping to find reinforcements. Do you risk him going after him to stop him from alerting the castle, even though that may mean exposing your party to surprise attacks or other encounters, or do you allow him to go free to alert the rest of the castle and bunker down? Or do you use the opportunity to escape and try again another time? Or, to take Jarmo's idea a bit further, you could have certain enemies attempt to surrender to you. The majority of enemies would either flee or fight to the death, but certain NPCs could yield to the party leading to some interesting interactions/options. You could also implement the mechanic into formations, where certain formations give a boost to morale and having a formation break because you're getting overrun/flanked intensifies the rate of morale loss.

I am not against romances in games, per se. I'm just against romances in this particular game. In the sense of "pick your own pixel" dynamics. If it's apart of the narrative and thematic core of course, then it is what it is. But otherwise, it seems like a waste of limited resources and unlike other certain studios, Obsidian doesn't need to hang their hat on that "pick your own pixel partner" dynamic to have "well written and deep" character interactions.

Completely agree with the premise of the OP. It'd also be nice if formations not only affected combat rolls, but had inherent strengths and weaknesses that make sense for the formation. Like how a shield wall reduces mobility, leaving them vulnerable to flanking and backstabbing Rogues. Or how magical AoEs would devastate a tightly grouped testudo. Seeing the AI utilise formations at times, and figuring out how to exploit said formations would also make for interesting encounters I think.