trie: Rewrite findPath without async-promise-executor

[ScottyPoi]

Followup to #2962 (shelved)

This takes the frist step towards simplifying and improving trie.findPath().

While the logic is fundamentally the same, this refactors findPath without the async-promise-executor API, and uses the async/await syntax directly to achieve the same result. The resolve and reject calls have been replaced with return and throw respectively.

[codecov]

Codecov Report

Merging #3015 (96154f9) into master (fdafe00) will increase coverage by 0.00%.
The diff coverage is 89.79%.

Additional details and impacted files

Flag	Coverage Δ
block	88.73% <ø> (ø)
blockchain	92.58% <ø> (ø)
client	87.34% <ø> (+0.06%)	⬆️
common	98.18% <ø> (ø)
ethash	?
evm	69.64% <ø> (ø)
rlp	?
statemanager	84.37% <ø> (ø)
trie	89.88% <89.79%> (-0.38%)	⬇️
tx	96.38% <ø> (ø)
util	86.78% <ø> (ø)
vm	74.66% <ø> (ø)

Flags with carried forward coverage won't be shown. Click here to find out more.

[ScottyPoi]

PR #3015 (commit: 96154f9)

-------
Benchmarking
-------
Benchmarking
1k-3-32-ran x 4 ops/sec @ 239ms/op
1k-5-32-ran x 3 ops/sec @ 263ms/op
1k-9-32-ran x 3 ops/sec @ 256ms/op
1k-1k-32-ran x 3 ops/sec @ 262ms/op
1k-1k-32-mir x 3 ops/sec @ 276ms/op
Checkpointing: 100 iterations x 14,556 ops/sec @ 68μs/op ± 12.43% (min: 43μs, max: 372μs)
Checkpointing: 500 iterations x 4,740 ops/sec @ 210μs/op ± 14.02% (min: 63μs, max: 3ms)
Checkpointing: 1000 iterations x 3,396 ops/sec @ 294μs/op ± 9.97% (min: 171μs, max: 4ms)
Checkpointing: 5000 iterations x 2,512 ops/sec @ 397μs/op ± 4.57% (min: 183μs, max: 8ms)
RAM: rss=598.0mb heap=504.4mb used=475.1mb ext=7.2mb arr=4.8mb
1k-3-32-ran x 0 ops/sec @ 3969ms/op
1k-5-32-ran x 3 ops/sec @ 327ms/op
1k-9-32-ran x 3 ops/sec @ 321ms/op
1k-1k-32-ran x 2 ops/sec @ 345ms/op
1k-1k-32-mir x 3 ops/sec @ 307ms/op
Checkpointing: 100 iterations x 10,548 ops/sec @ 94μs/op ± 31.51% (min: 56μs, max: 1ms)
Checkpointing: 500 iterations x 4,665 ops/sec @ 214μs/op ± 7.23% (min: 78μs, max: 2ms)
Checkpointing: 1000 iterations x 2,863 ops/sec @ 349μs/op ± 3.93% (min: 197μs, max: 2ms)
Checkpointing: 5000 iterations x 2,584 ops/sec @ 386μs/op ± 1.58% (min: 241μs, max: 2ms)
RAM: rss=701.0mb heap=555.8mb used=521.9mb ext=57.9mb arr=55.4mb

[acolytec3]

Results look great (and code looks good to me to). Any idea why there is such a performance gain on lower iterations?

[acolytec3]

Actually, I ran the benchmarks locally on both this branch and master and I can't reproduce the faster numbers, though maybe that's just my laptop. Regardless, it doesn't appear to perform any worse.

[ScottyPoi]

Results look great (and code looks good to me to). Any idea why there is such a performance gain on lower iterations?

Not really 🤷

[acolytec3]

Results look great (and code looks good to me to). Any idea why there is such a performance gain on lower iterations?

Not really 🤷

No worries. Was just curious if there was some performance benefit to reviewing the async promise executor.

[ScottyPoi]

@holgerd77

i want to make sure you also see / approve of this change

[holgerd77]

Nice, yeah, looks good! 🙏🙂👍