performance: Use push_unchecked when building indices for element take in fixedsizedlist take

[robert3005]

Signed-off-by: Robert Kruszewski github@robertk.io

[codspeed-hq]

CodSpeed Performance Report

Merging this PR will improve performance by 57.7%

_{Comparing rk/fixedsizedlisttake (9db9fcb) with develop (b5b492c)¹}

Summary

⚡ 23 improved benchmarks
✅ 1115 untouched benchmarks
⏩ 1265 skipped benchmarks²

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	chunked_bool_into_canonical[(1000, 10)]	63.9 µs	42.9 µs	+48.86%
⚡	Simulation	take_fsl_nullable_random[1024, 1000]	32.6 ms	22.6 ms	+44.57%
⚡	Simulation	take_fsl_nullable_random[16, 1000]	534 µs	399.8 µs	+33.58%
⚡	Simulation	take_fsl_nullable_random[1024, 100]	3.1 ms	2 ms	+54.46%
⚡	Simulation	take_fsl_nullable_random[16, 100]	141.4 µs	123.4 µs	+14.62%
⚡	Simulation	take_fsl_nullable_random[256, 100]	796.7 µs	525.3 µs	+51.67%
⚡	Simulation	take_fsl_nullable_random[64, 100]	234 µs	165.4 µs	+41.53%
⚡	Simulation	take_fsl_nullable_random[256, 1000]	7.2 ms	4.7 ms	+53.18%
⚡	Simulation	take_fsl_nullable_random[4096, 100]	12.4 ms	8.1 ms	+53.44%
⚡	Simulation	take_fsl_nullable_random[64, 1000]	1.9 ms	1.3 ms	+47.84%
⚡	Simulation	take_fsl_nullable_random[4096, 1000]	131 ms	90.8 ms	+44.32%
⚡	Simulation	take_fsl_random[64, 1000]	2 ms	1.3 ms	+53.38%
⚡	Simulation	take_fsl_random[1024, 1000]	34.4 ms	23 ms	+49.44%
⚡	Simulation	take_fsl_random[16, 1000]	516.2 µs	367.3 µs	+40.56%
⚡	Simulation	take_fsl_random[16, 100]	75.2 µs	59.5 µs	+26.34%
⚡	Simulation	take_fsl_random[1024, 100]	3.1 ms	2 ms	+57.17%
⚡	Simulation	take_fsl_random[4096, 1000]	138.2 ms	92.7 ms	+49.1%
⚡	Simulation	take_fsl_random[64, 100]	220.5 µs	151.5 µs	+45.6%
⚡	Simulation	take_fsl_random[256, 100]	804.8 µs	519.5 µs	+54.93%
⚡	Simulation	take_fsl_random[4096, 100]	12.7 ms	8.2 ms	+55.64%
...	...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

Footnotes

1. No successful run was found on develop (332fd06) during the generation of this report, so b5b492c was used instead as the comparison base. There might be some changes unrelated to this pull request in this report. ↩

2. 1265 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩

[connortsui20]

This is nice! I'm now convinced that if we migrated all of our builders to vectors we would have huge perf improvements everywhere since it makes these kinds of optimizations so much easier.