feat: gate 60 tuple impls behind 'more-tuple-impls' feature flag

[DaniPopes]

Puts #699 behind a feature flag. 60 impls is 3x slower to compile than 24, this is for only alloy-sol-types, so it does not include potential slowdowns downstream in users of sol! and such. It's at 6-6.5s on a clean build of forge.

This is technically a breaking change but having >24 fields in a struct is something that's already so niche that I really don't mind fixing compile times for everyone else.

$ hyperfine --warmup 1 --prepare "cargo clean -p alloy-sol-types" --parameter-list n 16,24,32,40,48,60,72 "./gen_macros.py {n} && cargo b -p alloy-sol-types"
Benchmark 1: ./gen_macros.py 16 && cargo b -p alloy-sol-types
  Time (mean ± σ):     755.8 ms ±   9.5 ms    [User: 782.3 ms, System: 118.2 ms]
  Range (min … max):   742.1 ms … 770.0 ms    10 runs
 
Benchmark 2: ./gen_macros.py 24 && cargo b -p alloy-sol-types
  Time (mean ± σ):      1.004 s ±  0.020 s    [User: 1.026 s, System: 0.121 s]
  Range (min … max):    0.987 s …  1.038 s    10 runs
 
Benchmark 3: ./gen_macros.py 32 && cargo b -p alloy-sol-types
  Time (mean ± σ):      1.344 s ±  0.023 s    [User: 1.347 s, System: 0.137 s]
  Range (min … max):    1.314 s …  1.382 s    10 runs
 
Benchmark 4: ./gen_macros.py 40 && cargo b -p alloy-sol-types
  Time (mean ± σ):      1.791 s ±  0.025 s    [User: 1.783 s, System: 0.146 s]
  Range (min … max):    1.751 s …  1.821 s    10 runs
 
Benchmark 5: ./gen_macros.py 48 && cargo b -p alloy-sol-types
  Time (mean ± σ):      2.320 s ±  0.033 s    [User: 2.301 s, System: 0.156 s]
  Range (min … max):    2.295 s …  2.378 s    10 runs
 
Benchmark 6: ./gen_macros.py 60 && cargo b -p alloy-sol-types
  Time (mean ± σ):      3.339 s ±  0.065 s    [User: 3.291 s, System: 0.173 s]
  Range (min … max):    3.198 s …  3.397 s    10 runs
 
Benchmark 7: ./gen_macros.py 72 && cargo b -p alloy-sol-types
  Time (mean ± σ):      4.609 s ±  0.135 s    [User: 4.503 s, System: 0.222 s]
  Range (min … max):    4.462 s …  4.934 s    10 runs
 
Summary
  ./gen_macros.py 16 && cargo b -p alloy-sol-types ran
    1.33 ± 0.03 times faster than ./gen_macros.py 24 && cargo b -p alloy-sol-types
    1.78 ± 0.04 times faster than ./gen_macros.py 32 && cargo b -p alloy-sol-types
    2.37 ± 0.04 times faster than ./gen_macros.py 40 && cargo b -p alloy-sol-types
    3.07 ± 0.06 times faster than ./gen_macros.py 48 && cargo b -p alloy-sol-types
    4.42 ± 0.10 times faster than ./gen_macros.py 60 && cargo b -p alloy-sol-types
    6.10 ± 0.19 times faster than ./gen_macros.py 72 && cargo b -p alloy-sol-types

[onbjerg]

makes sense to me, do we know who would be affected by this?

[DaniPopes]

Don't know of any besides original PR author. cc @MatheoGrivault
My initial count of 24 was based on a basic analysis of all structs on GitHub searched with "lang: Solidity".