http: cache parser callbacks

[mcollina]

Two low-level reductions in the HTTP parser's per-request work, continuing the server hot-path cleanup from #65732, #65749 and #65802.

Cache the per-message callback lookups. on_headers_complete, on_body and on_message_complete do an object()->Get() for their JS callback (kOnHeadersComplete / kOnBody / kOnMessageComplete) on every message. Those callbacks are assigned once when the parser object is created (in the parsers FreeList factory) and never change afterwards, so the lookup is cached per parser in a v8::Global. The cache is cleared in Init() so a parser object that is re-initialized with different callbacks — which direct users of the binding do — stays correct.

Intern header field-name strings. When building the header array, field names are created with NewStringType::kInternalized instead of kNormal. Header field names repeat heavily across requests on a keep-alive connection (Host, User-Agent, Accept, …), so interning lets V8's string table return the same string object for each rather than allocating a fresh one per request. Header values are unique per request and stay non-interned.

No observable behavior change: field names keep their original case in rawHeaders, duplicate/unknown/mixed-case header handling is unchanged, and the full test-http-*, test-http2-*, test-https-*, test-net-*, test-stream-* and async-hooks suites pass (1254 tests), plus AsyncLocalStorage propagation is unaffected.

Benchmark (i7-7700, server pinned to one core, wrk -t2 -c50 on separate physical cores, CPU from /proc/<pid>/stat, 5 interleaved rounds):

workload	main	this PR
hello world, 1 request header	~24.9 µs/req	~24.4 µs/req
hello world, 8 request headers	~28.5 µs/req	~28.0 µs/req (lower in all 5 rounds)

A small (~2%) but consistent reduction.

—-

AI generated, humanly reviewed.

[nodejs-github-bot]

Review requested:

• @nodejs/http
• @nodejs/net

[mcollina]

@nodejs/diagnostics ptal. Not sure if anyone is monkeypatching those callbacks on the fly.

[nodejs-github-bot]

CI: https://ci.nodejs.org/job/node-test-pull-request/77666/

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.28%. Comparing base (bf9605d) to head (781531a).
⚠️ Report is 26 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff            @@
##             main   #66152    +/-   ##
========================================
  Coverage   90.28%   90.28%            
========================================
  Files         790      790            
  Lines      271642   272030   +388     
  Branches    51853    51934    +81     
========================================
+ Hits       245241   245612   +371     
+ Misses      16918    16906    -12     
- Partials     9483     9512    +29     

Files with missing lines	Coverage Δ
src/node_http_parser.cc	85.32% <100.00%> (+0.15%)	⬆️

... and 45 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.