src: use validate_ascii_with_errors for buffer.isAscii

[ChALkeR]

Tracking: #61041

Don't expect buffer.isAscii common-case to be ASCII, as that's exposed as a public API for checks.
The error version has better performance on non-ascii as it returns early, which could easily be e.g. >10x.

Refs: #46271 (comment),

But somehow the _errors version is also faster on ASCII, which was supposed to be a common case for validate_ascii?
I'm not sure what's up, might be a simdutf bug? cc @lemire

On another note: v8 is also likely using validate_ascii incorrectly

buffer.isAscii perf:

main:

Test	Size	Throughput	Mean Time
Latin lipsum (ASCII)	84.902 KiB	105.08 GiB/s	0.001 ms
Arabic lipsum	79.771 KiB	103.88 GiB/s	0.001 ms
Chinese lipsum	68.203 KiB	101.78 GiB/s	0.001 ms

PR

Test	Size	Throughput	Mean Time
Latin lipsum (ASCII)	84.902 KiB	117.97 GiB/s	0.001 ms
Arabic lipsum	79.771 KiB	1836.47 GiB/s	0.000 ms
Chinese lipsum	68.203 KiB	1564.76 GiB/s	0.000 ms

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 88.52%. Comparing base (4f24aff) to head (e78cdd1).
⚠️ Report is 9 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #61122      +/-   ##
==========================================
- Coverage   88.53%   88.52%   -0.02%     
==========================================
  Files         703      703              
  Lines      208546   208547       +1     
  Branches    40217    40217              
==========================================
- Hits       184634   184607      -27     
- Misses      15926    15949      +23     
- Partials     7986     7991       +5     

Files with missing lines	Coverage Δ
src/node_buffer.cc	68.90% <100.00%> (+0.03%)	⬆️

... and 46 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.

[lemire]

Thanks for the ping.