Rewrite Array#each in Ruby

[k0kubun]

Similarly to Integer#times #8388, this PR rewrites Array#each in Ruby.

microbenchmark

Interpreter

It's a bit slower than the original C method.

$ benchmark-driver benchmark/loop_each.yml -v --chruby 'before;after'
before: ruby 3.4.0dev (2024-01-13T00:28:26Z master f7178045bb) [x86_64-linux]
after: ruby 3.4.0dev (2024-01-13T04:31:50Z builtin-array-each 18c9a45314) [x86_64-linux]
Warming up --------------------------------------
           loop_each      2.129 i/s -       3.000 times in 1.408852s (469.62ms/i)
Calculating -------------------------------------
                         before       after
           loop_each      2.103       1.545 i/s -       6.000 times in 2.852494s 3.882698s

Comparison:
                        loop_each
              before:         2.1 i/s
               after:         1.5 i/s - 1.36x  slower

YJIT

It's 5.3x faster.

$ benchmark-driver benchmark/loop_each.yml -v --chruby 'before::before --yjit-call-threshold=1;after::after --yjit-call-threshold=1'
before: ruby 3.4.0dev (2024-01-13T00:28:26Z master f7178045bb) +YJIT [x86_64-linux]
after: ruby 3.4.0dev (2024-01-13T04:31:50Z builtin-array-each 18c9a45314) +YJIT [x86_64-linux]
Warming up --------------------------------------
           loop_each      2.451 i/s -       3.000 times in 1.223915s (407.97ms/i)
Calculating -------------------------------------
                         before       after
           loop_each      2.456      13.074 i/s -       7.000 times in 2.850521s 0.535404s

Comparison:
                        loop_each
               after:        13.1 i/s
              before:         2.5 i/s - 5.32x  slower

[k0kubun]

Given https://bugs.ruby-lang.org/issues/20182#note-5, #9533 will take this over.

[alfuken]

@k0kubun forgive me my ignorance, but just out of curiosity: why is there a change in tests from each to reverse_each?.. If this PR is a refactoring of internal mechanism, shouldn't tests remain the same, to ensure (non-)regression integrity?..

[k0kubun]

The change of this line was actually not necessary, so I reverted it in the end #9533 (comment).