Implement latin1 encoded JsStrings

[HalidOdat]

This fixes #3097

This memory optimization is implemented by both spidermonkey and v8.

This PR encodes strings that can be represented as latin1 as a byte array, instead of u16 array, this cuts latin1 strings size by 2x.

This may also have some interesting optimization that could be applied when we know that the string is in the ASCII range. Decided to not preserve asciiness though it can work with asciiness preserved it would complicate other string optimizations. AFIK v8 does not preserve the asciiness. Preserving the asciiness could lead to some interesting optimizations but at the cost of a lot of complexity which from my testing doesn't seem to be worth it.

[github-actions]

Test262 conformance changes

Test result	main count	PR count	difference
Total	50,731	50,731	0
Passed	42,973	42,973	0
Ignored	1,395	1,395	0
Failed	6,363	6,363	0
Panics	18	18	0
Conformance	84.71%	84.71%	0.00%

[codecov]

Codecov Report

Attention: Patch coverage is 62.01084% with 631 lines in your changes are missing coverage. Please review.

Project coverage is 44.47%. Comparing base (6ddc2b4) to head (52d192a).
Report is 148 commits behind head on main.

❗ Current head 52d192a differs from pull request most recent head bc980e4. Consider uploading reports for the commit bc980e4 to get more accurate results

Files	Patch %	Lines
boa_cli/src/debug/string.rs	0.00%	47 Missing ⚠️
boa_engine/src/builtins/string/mod.rs	66.41%	44 Missing ⚠️
boa_engine/src/string/mod.rs	79.27%	40 Missing ⚠️
boa_engine/src/builtins/temporal/calendar/mod.rs	47.76%	35 Missing ⚠️
boa_engine/src/string/slice.rs	65.11%	30 Missing ⚠️
boa_engine/src/builtins/intl/collator/mod.rs	10.00%	27 Missing ⚠️
boa_engine/src/builtins/regexp/mod.rs	66.66%	27 Missing ⚠️
boa_engine/src/string/str.rs	52.08%	23 Missing ⚠️
boa_engine/src/builtins/intl/segmenter/mod.rs	15.00%	17 Missing ⚠️
boa_engine/src/builtins/intl/plural_rules/mod.rs	15.78%	16 Missing ⚠️
... and 75 more

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3450      +/-   ##
==========================================
- Coverage   47.24%   44.47%   -2.77%     
==========================================
  Files         476      490      +14     
  Lines       46892    50360    +3468     
==========================================
+ Hits        22154    22398     +244     
- Misses      24738    27962    +3224     

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

[HalidOdat]

This is ready for review/merge 🥳

In terms of performance it's pretty much the same as on main, a very slight regression that doesn't effect the overall score.

I did some analysis on what is the percentage of latin1 and u16 through out the combined.js execution for allocation strings (so static strings are not counted) and found that ~99.63 are latin1 strings, some methods still converts to u16, we could reduce this number even more by not eagerly converting to u16.

latin1: 9459373,
latin1_size: 64732594 (in bytes),
u16: 34755,
u16_size: 4346430 (in bytes),

There was also a reduction on binary size by ~64KB!

Checked locally and there is not difference between main and this PR in conformance, it seems that data repo's test262 data is not being updated?

[jedel1043]

Checked locally and there is not difference between main and this PR in conformance, it seems that data repo's test262 data is not being updated?

That's weird. I checked the output of more recent PRs such as dependabot's and everything looks fine.

[HalidOdat]

That's weird. I checked the output of more recent PRs such as dependabot's and everything looks fine.

I reran the test262 workflow and now it shows the correct conformance! :)

[raskad]

Just got a few comments, otherwise looks great!

[jedel1043]

Really impressive work!