Reuse copy buffers in OCaml rewatch

[cknitt]

OCaml rewatch allocated a fresh 64 KiB buffer for every file it published (.cmi, .cmj, .cmt, ASTs, sources), and two more for every same-sized .cmi it compared. An allocation profile (Gc.Memprof) of a clean tests/tests build showed these buffers as the largest single source of allocation: 17.5% of all words. Blocks that size go straight to the major heap, so they also drive major-GC work, which with several domains stops all of them.

Each domain now keeps one copy buffer and one comparison buffer and reuses them.

Clean builds of tests/tests without string_unicode_test (see #8866), idle Linux VM with 10 cores, median of 7 interleaved runs:

	Before	After
Allocated words	241M	198.5M (−18%)
Wall, 1 domain	1.312 s	1.246 s (−5%)
Wall, 8 domains	0.422 s	0.370 s (−12%)

🤖 Generated with Claude Code

[pkg-pr-new]

Open in StackBlitz

rescript

npm i https://pkg.pr.new/rescript@8869

@rescript/belt

npm i https://pkg.pr.new/@rescript/belt@8869

@rescript/darwin-arm64

npm i https://pkg.pr.new/@rescript/darwin-arm64@8869

@rescript/darwin-x64

npm i https://pkg.pr.new/@rescript/darwin-x64@8869

@rescript/linux-arm64

npm i https://pkg.pr.new/@rescript/linux-arm64@8869

@rescript/linux-x64

npm i https://pkg.pr.new/@rescript/linux-x64@8869

@rescript/runtime

npm i https://pkg.pr.new/@rescript/runtime@8869

@rescript/win32-x64

npm i https://pkg.pr.new/@rescript/win32-x64@8869

commit: d9d9d32

[github-actions]

Developer playground preview: https://rescript-lang.github.io/rescript/dev-playground/?version=pr-8869