If you could lose a capital ship and still come out significantly ahead from a bounty, then winning without loses would be absurdly profitable. Combat isn't built around the notion that players can't lose ships, it's built around the notion that the player should minimize loses to maximize profits. I don't see what's wrong with that, you get rewarded for making better decisions. I think it's important to make sure that you don't have to play in un-fun ways to avoid loses. The player can always choose to take easier or harder fights if they want more or less of a challenge.
Agreed. I'm assuming the unfun way you are speaking of is the ship losses outside of the player's tactical decisions? Correct me if that is not the case.
Overall, I think the solution to bounties is more about bounty mechanics and scale. Losing a capital is fine if the scale of your campaign situation allows it through monthly income. If you must defeat four other capitals in the bounty, however, and losing one negates your profits then you won't likely take the mission in the first place unless you have equal or more capitals in your fleet, you have built your fleet around such a bounty and have extensive knowledge of combat balance, or you have some other element/strategy that allows you to
know you won't lose a capital in this situation. Alternatively, you could have some other kind of campaign incentive to take the mission and take a net credit loss in the process.
FWIW I'm not saying that your concerns regarding the spawn location knowledge that AI ships would need is wrong or wouldn't be a nice addition. It's more that I would prioritize less need for keeping ships alive in combat using existing mechanics like the ship recovery system since that is more realistic from an AI edge-case perspective or simply inexperienced play or bad luck with positioning or other RNG components. Making the AI feel perfect in every situation is a very difficult thing to do.
What I really don't like is when the AI kills itself in ways that don't reflect the decisions you've made. The AI should do as well as it can in light of your decisions, it shouldn't do poorly in spite of your decisions. The AI should know basic game mechanics like spawn areas. My concern is that the things the player has to do to prevent the AI from killing itself in stupid ways due to weird flanks will not be fun. If the AI is adjusted to better deal with these situations, I think the OP's suggestion could work.
I think this is partially true and I also think that anyone would agree that when you make a decision as the player and the AI seems to not respond well it doesn't feel good. However, I also think that flanking frigates aren't necessarily a recipe for that feeling. To be fair, it does promote more situational awareness from a tactics perspective. What I like about the idea is that it gives a concrete use for smaller ships in the late game in a way that tactically makes sense to me. Adjustments such as you suggest might also be a requirement and that would increase the amount of effort it would require, but I don't think it's a bad idea as whole.
As to the impact it would have on combat? It might make early advantages in the beginning of a battle less likely to snowball. I think that scenario could also result in battles taking too long to conclude but considering the fleet size/number changes this concern is less likely to be the case in the next update.