refactor: simplify async op error construction

[crowlKats]

Removes usage of serde_v8 for constructing errors

[littledivy]

Can you run a op_void_async_deferred microbenchmark? Slighty worried this will cause a regression because it's in the hot path. Maybe { resolve, reject } is faster because of shape optimizations.

[crowlKats]

main:

test bench_op_async_void_deferred         ... bench:     739,054 ns/iter (+/- 77,089)
test bench_op_async_void_deferred_nofast  ... bench:     715,870 ns/iter (+/- 78,892)
test bench_op_async_void_deferred_return  ... bench:     678,101 ns/iter (+/- 38,345)

PR:

test bench_op_async_void_deferred         ... bench:     712,190 ns/iter (+/- 61,168)
test bench_op_async_void_deferred_nofast  ... bench:     742,969 ns/iter (+/- 52,853)
test bench_op_async_void_deferred_return  ... bench:     689,404 ns/iter (+/- 41,140)

PR with suggestion:

test bench_op_async_void_deferred         ... bench:     742,152 ns/iter (+/- 66,355)
test bench_op_async_void_deferred_nofast  ... bench:     736,991 ns/iter (+/- 70,942)
test bench_op_async_void_deferred_return  ... bench:     689,867 ns/iter (+/- 39,801)

[marvinhagemeister]

If the array allocation is a concern we could also move two arrays of ids instead:

const promiseRing = ArrayPrototypeFill(new Array(RING_SIZE), NO_PROMISE);
const rejectRing = ArrayPrototypeFill(new Array(RING_SIZE), NO_PROMISE);

// later
const promise = new Promise((resolve, reject) => {
  promiseRing[idx] = resolve;
  promiseReject[idx] = reject;
});

Basically keep two rings at the same time. That should be better for memory and GC pressure.

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 56.00000% with 66 lines in your changes missing coverage. Please review.

Project coverage is 81.66%. Comparing base (0c7f83e) to head (c5020a1).
Report is 229 commits behind head on main.

Files with missing lines	Patch %	Lines
core/ops_builtin.rs	42.64%	39 Missing ⚠️
core/ops_builtin_v8.rs	31.25%	22 Missing ⚠️
core/runtime/op_driver/mod.rs	25.00%	3 Missing ⚠️
core/runtime/op_driver/op_results.rs	85.71%	1 Missing ⚠️
testing/checkin/runner/ops_worker.rs	75.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1039      +/-   ##
==========================================
+ Coverage   81.43%   81.66%   +0.22%     
==========================================
  Files          97      102       +5     
  Lines       23877    27927    +4050     
==========================================
+ Hits        19445    22806    +3361     
- Misses       4432     5121     +689     

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

[littledivy]

LGTM