perf(Game/IGame): speedup multiplication theorems

[plp127]

Speedup multiplication theorems in Game/IGame

I measured

• private theorem mul_comm' goes from 1500ms to 50ms
• private theorem neg_mul' goes from 1000ms to 220ms

[plp127]

The private theorem zero_mul' is 80ms, so that one is fine, but private theorem one_mul' is 3000ms. I didn't touch one_mul' because the proof is so short, but should I try speeding that up too?

[vihdzp]

I'd personally feel sad replacing a succinct proof for a fifteen-line case bash in favor of like two seconds of build-time. But if you can find terse proofs (or proofs that are no worse) that are faster, then that's obviously a net positive.

[vihdzp]

I'm not all too sure how I feel about mul_eq'' having a "manual proof", to be honest. I'll leave the decision of merging this up to @tristan-f-r.