AI: include ProduceMana in the performance-mode replacement zone skip

[liamiak]

getReplacementList visits every card in the game. On the mana hot path that means rebuilding the replacement-effect list of ~1700 cards to find the handful that could ever apply — and it runs per mana source, per AI cost check.

The performance-mode zone skip added in #11160 already covers Tap/Untap, on the grounds that those replacements are only active from the battlefield or the command zone. This extends the same skip to ProduceMana, which groupSourcesByManaColor checks per mana ability on top of the Tap checks.

Why ProduceMana qualifies

Survey of forge-gui/res/cardsfolder:

• 25 Event$ ProduceMana replacement-effect lines
• 23 declare ActiveZones — every one is Battlefield and/or Command
• the 2 that don't (False Dawn, Quarum Trench Gnomes) are both SP$/AB$ Effect | ReplacementEffects$ ..., i.e. hosted on Effect cards, which live in the command zone

So no ProduceMana replacement effect in the card database is active outside battlefield/command — the same justification that licensed Tap/Untap.

Measurements

3 headless 4-player games, "Big 240531" mirror. Normalised per call, since sims aren't deterministic and game length varies between runs:

	before	after
Tap scan	2788.6 µs	194.4 µs
ProduceMana scan	2749.2 µs	178.7 µs
RE-list rebuilds per scan	1681	34

Tap was already covered by #11160; ProduceMana is what this change adds. The traversal itself is unchanged — this only avoids the per-card getReplacementEffects() rebuild, which measured as ~78% of the scan cost.

Note

While measuring this I found that the skip never fires by default: PERFORMANCE_MODE defaults to false, so #11160's optimisation is inert for most users and is not exercised by CI. That's a separate discussion and I'll open it separately rather than bundle it here.

Edit/note from actual human: It seems claude (opus 4.8) is missing the note on identifying AI-generated code in this context window, so I'm adding it manually. We're both trying to find the right balance of small PRs vs optimization wins, and we're both having fun (for likely quite different values of fun) getting our head around the engine, and the norms desired by the forge community - hope this help, and happy to shift if this is too much of a pain/distraction from the eternal grind of new sets, but also thrilled to maybe be able to help with something that's brought me much joy!

[tool4ever]

ok as temporary workaround