Make QRFs and singleAttacks behave semi-reasonably with maxUnits

[jaj22]

What type of PR is this.

1. Bug
2. Change
3. Enhancement

What have you changed and why?

Support QRFs and singleAttack (marker recapture counterattack) were using a maxUnits check (through chooseAttackType) that didn't account for unit locality. In Antistasi, maxUnits applies per machine rather than globally, so the permitted unit count scales up with the number of headless clients. As a result, QRFs and singleAttacks will have been much rarer or smaller than intended when headless clients were used.

This PR fixes the maxUnits checks so that they ignore units on other machines. A buffer is left for QRFs so that wavedCA doesn't get completely crippled by running on the same machine.

Please specify which Issue this PR Resolves.

closes #2098

Please verify the following and ensure all checks are completed.

1. Have you loaded the mission in LAN host?
2. Have you loaded the mission on a dedicated server?

Is further testing or are further changes required?

1. No
2. Yes (Please provide further detail below.)

maxUnits behaviour is generally terrible but that's a bigger problem to be addressed later.

More problematic is that singleAttacks and QRFs likely didn't spawn at all on the community servers, so that would invalidate any balance data. Might need to buff up economicsAI again, for example, and the upper-end vehicle counts in singleAttacks and QRFs are likely far too high, as that end hasn't been tested in an appropriate case.

How can the changes be tested?

WIP

[jaj22]

Had a better idea. Make maxUnits a global value and don't attempt to micro-manage individual machines. Should make things more stable and transparent.

[HakonRydland]

Had a better idea. Make maxUnits a global value and don't attempt to micro-manage individual machines. Should make things more stable and transparent.

couldnt that cause load issues where say one machine could have most of the units on it and another having like 10 units?

[jaj22]

couldnt that cause load issues where say one machine could have most of the units on it and another having like 10 units?

Routines that use the HCs (garrisons, missions, singleAttack, reinfPatrols) are sent to the one with the lowest unit count anyway. Major attacks and supports (including post-2.4 QRFs) run on the server, but as there's only one attack permitted at once, and QRFs are limited by other means (this bit is questionable), I don't think there's much risk of heavy asymmetry. The worst case is already happening: Single HC gets overloaded with garrisons and singleAttacks while the server is basically idle.

[jaj22]

On second thoughts, the worst part is that garrisons ignore the limit but count towards it, potentially blocking out everything else. This case is potentially worse with a global limit because at least garrisons don't spawn on the server when you have HCs, so QRFs and major attacks should still run.

Maybe leave this as-is and rethink the whole system later. Might throw in another PR to sort out some particularly spammy garrisons though.