bench: lazily initialize script benchmarks

[iammdzaidalam]

Closes #5169

Summary

Defer script benchmark setup until the selected benchmark actually runs.

benches/benches/scripts.rs was eagerly reading, parsing, compiling, and evaluating every script during registration, so filtered runs could still fail on unrelated entries before reaching the requested benchmark.

This moves that setup behind the benchmark closure and caches the prepared state per benchmark, so unmatched scripts are no longer initialized.

Changes

• add a small PreparedScriptBench helper for cached per-benchmark state
• move script file reading into lazy setup
• move Context creation, runtime registration, parse/compile/evaluate, and main lookup into lazy setup
• keep benchmark discovery and existing v8-benches group config unchanged
• cache the prepared script once per matched benchmark so setup is not repeated during measurement

Verification

Ran locally:

• cargo fmt --check
• cargo check -p boa_benches
• cargo bench -p boa_benches -- --list
• cargo bench -p boa_benches -- call-loop

Also temporarily added logging in the lazy init path to verify behavior:

• call-loop only initialized basic/call-loop.js
• a nonexistent filter initialized nothing
• deltablue only initialized v8-benches/deltablue.js

So filtered runs no longer initialize unrelated scripts first.

[github-actions]

Test262 conformance changes

Test result	main count	PR count	difference
Total	52,963	52,963	0
Passed	50,126	50,126	0
Ignored	2,025	2,025	0
Failed	812	812	0
Panics	0	0	0
Conformance	94.64%	94.64%	0.00%

Tested main commit: b8c684580787968c613045c2588834b4442af518
Tested PR commit: eb961ba832f27bd8fec14127a20a2f81499d7d2d
Compare commits: b8c6845...eb961ba

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 59.80%. Comparing base (6ddc2b4) to head (eb961ba).
⚠️ Report is 906 commits behind head on main.

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #5190       +/-   ##
===========================================
+ Coverage   47.24%   59.80%   +12.55%     
===========================================
  Files         476      582      +106     
  Lines       46892    63414    +16522     
===========================================
+ Hits        22154    37923    +15769     
- Misses      24738    25491      +753     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 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.

[jedel1043]

I don't think we should put the initialization code inside the benchmark, it'll just pollute the results.

[hansl]

Was about to say. The reason we use the main function is to benchmark specific bits of VM, not the initialization and parsing (and optimization, etc).

[iammdzaidalam]

Oh, I was mainly trying to avoid the eager init issue but putting it inside the benchmark isnt the right tradeoff here... thinking of instead filtering before registration so only matching scripts get initialized, and keeping the setup outside bench_function like before

does that sound like the right direction?