Remove direct conversion from &str to JsValue/PropertyKey.

[jedel1043]

This is with the objective of avoiding hidden conversions while doing the conversion from a string literal to a JsValue. Removing the direct conversion to JsValue makes it a bit easier to realize that this is doing a costly conversion of
string literal -> (runtime) decode UTF-8 to UTF-16 -> (runtime) JsString -> (runtime) JsValue
and pushes the alternative of using the js_string macro that makes the optimal conversion
string literal -> (compile time) &[u16] -> (runtime) JsString -> (runtime) JsValue

This should technically improve realm initialization times, but let's see what the benchmarks show.

[github-actions]

Test262 conformance changes

Test result	main count	PR count	difference
Total	95,574	95,574	0
Passed	75,193	75,193	0
Ignored	19,482	19,482	0
Failed	899	899	0
Panics	0	0	0
Conformance	78.68%	78.68%	0.00%

[codecov]

Codecov Report

Attention: 260 lines in your changes are missing coverage. Please review.

Comparison is base (615ae4e) 49.69% compared to head (218c53c) 49.72%.
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3319      +/-   ##
==========================================
+ Coverage   49.69%   49.72%   +0.03%     
==========================================
  Files         443      443              
  Lines       43167    43378     +211     
==========================================
+ Hits        21451    21569     +118     
- Misses      21716    21809      +93     

Files	Coverage Δ
boa_engine/src/builtins/array/array_iterator.rs	81.13% <100.00%> (+0.36%)	⬆️
boa_engine/src/builtins/array/mod.rs	77.16% <100.00%> (+0.06%)	⬆️
boa_engine/src/builtins/array_buffer/mod.rs	10.56% <100.00%> (+0.63%)	⬆️
boa_engine/src/builtins/async_function/mod.rs	39.13% <100.00%> (ø)
boa_engine/src/builtins/async_generator/mod.rs	30.57% <100.00%> (ø)
...ngine/src/builtins/async_generator_function/mod.rs	46.15% <100.00%> (ø)
boa_engine/src/builtins/dataview/mod.rs	12.63% <100.00%> (+1.46%)	⬆️
boa_engine/src/builtins/error/aggregate.rs	22.22% <100.00%> (ø)
boa_engine/src/builtins/error/eval.rs	72.72% <100.00%> (ø)
boa_engine/src/builtins/error/mod.rs	89.58% <100.00%> (ø)
... and 97 more

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

[github-actions]

Benchmark for 69d2900

Click to view benchmark

Test	Base	PR	%
Arithmetic operations (Compiler)	6.5±0.05ns	7.1±0.00ns	+9.23%
Arithmetic operations (Execution)	435.5±1.63ns	440.7±0.78ns	+1.19%
Arithmetic operations (Parser)	9.1±0.11µs	8.9±0.23µs	-2.20%
Array access (Compiler)	6.5±0.01ns	7.1±0.02ns	+9.23%
Array access (Execution)	6.7±14.79µs	6.6±14.15µs	-1.49%
Array access (Parser)	15.8±0.32µs	15.9±0.32µs	+0.63%
Array creation (Compiler)	6.5±0.04ns	7.1±0.01ns	+9.23%
Array creation (Execution)	863.9±1247.58µs	873.4±1422.13µs	+1.10%
Array creation (Parser)	19.1±0.36µs	19.7±0.12µs	+3.14%
Array pop (Compiler)	6.5±0.02ns	7.1±0.01ns	+9.23%
Array pop (Execution)	285.7±256.58µs	287.2±286.76µs	+0.53%
Array pop (Parser)	179.7±5.02µs	177.9±2.06µs	-1.00%
Boolean Object Access (Compiler)	6.5±0.01ns	7.1±0.00ns	+9.23%
Boolean Object Access (Execution)	8.4±33.20µs	5.7±11.15µs	-32.14%
Boolean Object Access (Parser)	17.7±0.38µs	18.2±0.08µs	+2.82%
Clean js (Compiler)	6.5±0.02ns	7.1±0.02ns	+9.23%
Clean js (Execution)	571.3±546.20µs	569.0±445.77µs	-0.40%
Clean js (Parser)	46.4±5.43µs	45.0±4.26µs	-3.02%
Create Realm	660.2±1988.85µs	581.5±1881.54µs	-11.92%
Dynamic Object Property Access (Compiler)	6.5±0.08ns	7.1±0.03ns	+9.23%
Dynamic Object Property Access (Execution)	5.5±24.29µs	3.7±6.91µs	-32.73%
Dynamic Object Property Access (Parser)	13.9±0.30µs	14.4±0.04µs	+3.60%
Fibonacci (Compiler)	6.5±0.01ns	7.1±0.00ns	+9.23%
Fibonacci (Execution)	864.9±1286.32µs	1385.5±6049.42µs	+60.19%
Fibonacci (Parser)	26.3±44.04µs	22.9±0.13µs	-12.93%
For loop (Compiler)	6.5±0.05ns	7.1±0.02ns	+9.23%
For loop (Execution)	22.6±59.85µs	62.4±391.19µs	+176.11%
For loop (Parser)	20.2±1.60µs	21.0±0.11µs	+3.96%
Mini js (Compiler)	6.5±0.03ns	7.1±0.01ns	+9.23%
Mini js (Execution)	491.0±1.68µs	496.4±2.90µs	+1.10%
Mini js (Parser)	41.3±5.10µs	40.7±6.07µs	-1.45%
Number Object Access (Compiler)	6.5±0.02ns	7.1±0.00ns	+9.23%
Number Object Access (Execution)	4.3±7.23µs	4.1±6.39µs	-4.65%
Number Object Access (Parser)	14.5±0.02µs	14.5±0.31µs	0.00%
Object Creation (Compiler)	6.7±0.04ns	7.1±0.02ns	+5.97%
Object Creation (Execution)	4.4±16.99µs	3.2±9.41µs	-27.27%
Object Creation (Parser)	12.2±0.21µs	11.9±0.34µs	-2.46%
RegExp (Compiler)	6.5±0.01ns	7.1±0.01ns	+9.23%
RegExp (Execution)	17.9±49.48µs	14.7±19.89µs	-17.88%
RegExp (Parser)	13.2±0.31µs	12.7±0.36µs	-3.79%
RegExp Creation (Compiler)	6.5±0.15ns	7.1±0.01ns	+9.23%
RegExp Creation (Execution)	19.5±126.77µs	7.7±13.45µs	-60.51%
RegExp Creation (Parser)	10.8±0.14µs	11.4±0.18µs	+5.56%
RegExp Literal (Compiler)	6.5±0.01ns	7.1±0.03ns	+9.23%
RegExp Literal (Execution)	14.8±19.98µs	12.7±13.53µs	-14.19%
RegExp Literal (Parser)	20.8±58.46µs	14.9±0.10µs	-28.37%
RegExp Literal Creation (Compiler)	6.5±0.01ns	7.1±0.02ns	+9.23%
RegExp Literal Creation (Execution)	8.7±10.12µs	6.5±7.73µs	-25.29%
RegExp Literal Creation (Parser)	12.8±0.22µs	12.7±0.16µs	-0.78%
Static Object Property Access (Compiler)	6.6±0.29ns	7.1±0.00ns	+7.58%
Static Object Property Access (Execution)	3.2±7.28µs	3.3±7.61µs	+3.12%
Static Object Property Access (Parser)	12.8±0.27µs	13.0±0.28µs	+1.56%
String Object Access (Compiler)	6.5±0.01ns	7.1±0.03ns	+9.23%
String Object Access (Execution)	6.7±9.49µs	10.2±38.93µs	+52.24%
String Object Access (Parser)	17.4±0.61µs	17.7±0.03µs	+1.72%
String comparison (Compiler)	6.6±0.00ns	7.1±0.02ns	+7.58%
String comparison (Execution)	3.1±7.58µs	3.5±9.99µs	+12.90%
String comparison (Parser)	18.5±4.47µs	17.1±2.56µs	-7.57%
String concatenation (Compiler)	6.5±0.01ns	7.1±0.07ns	+9.23%
String concatenation (Execution)	2.5±5.71µs	2.7±6.69µs	+8.00%
String concatenation (Parser)	11.6±1.93µs	10.9±0.35µs	-6.03%
String copy (Compiler)	6.5±0.05ns	7.1±0.00ns	+9.23%
String copy (Execution)	2.5±6.51µs	2.8±8.20µs	+12.00%
String copy (Parser)	7.7±0.18µs	30.3±215.00µs	+293.51%
Symbols (Compiler)	6.6±0.06ns	7.2±0.28ns	+9.09%
Symbols (Execution)	1265.4±585.65ns	1199.1±107.76ns	-5.24%
Symbols (Parser)	6.1±0.16µs	6.0±0.14µs	-1.64%

[jedel1043]

Ok, results are a bit erratic (+293.51% in parsing string copy without even touching boa_parser???), so I'll run them locally.

[jedel1043]

Yep, some nice wins by skipping the conversions.
While main uses 600KB of data:

This PR uses 573KB:

Mainly because conversions use an additional 24KB of intermediary allocations:

It would be cool if we could also remove the additional 12KB used by the thread local cache of static strings (maybe a phf?):

[raskad]

I really like the change. The cost of conversions like this should be very clear.

[nekevss]

I like this! I am a bit curious about documentation on using js_string over &str for contributors and anybody consuming the API. But I'm fine with merging as is.