Run benchmark script once

[dapplion]

Motivation

We must not run benchmark script twice:

• It won't persist all benchmark data but only the data associated to the last package
• Therefore there won't be any data to compare for all packages but the last

Description

• Run benchmark runner once with a glob that grabs perf tests from all packages, without lerna.
• Simplify BLS initialization to work with this approach
• Run benchmark Github action not in forks
• Disable unnecessary benchmarks in CI
• Run CheckpointStateCache initialization in before block
• Add basic BLS performance test

[codeclimate]

Code Climate has analyzed commit 73e777b and detected 0 issues on this pull request.

View more on Code Climate.

[github-actions]

Performance Report

✔️ no performance regression detected

🚀🚀 Significant benchmark improvement detected

Benchmark suite	Current: 89c57b9	Previous: bad3f6e	Ratio
Process block - 250000 vs - with 0 validator exit	224.98 us/op	5.7165 ms/op	0.04
epoch - 250000 vs - processSlashings	33.791 us/op	3.5590 ms/op	0.01
shuffle list - 512 els	144.20 us/op	1.4239 ms/op	0.10
shuffle list - 16384 els	1.8196 ms/op	15.937 ms/op	0.11
shuffle list - 250000 els	24.698 ms/op	225.32 ms/op	0.11

Full benchmark results

Benchmark suite	Current: 89c57b9	Previous: bad3f6e	Ratio
Process block - 250000 vs - with 0 validator exit	224.98 us/op	5.7165 ms/op	0.04
Process block - 250000 vs - with 1 validator exit	20.519 ms/op	25.965 ms/op	0.79
Process block - 250000 vs - with 16 validator exits	21.978 ms/op	26.717 ms/op	0.82
epoch - 250000 vs - processJustificationAndFinalization	69.331 us/op	119.72 us/op	0.58
epoch - 250000 vs - processRewardsAndPenalties	275.41 ms/op	359.30 ms/op	0.77
epoch - 250000 vs - processRegistryUpdates	9.9640 us/op	10.317 us/op	0.97
epoch - 250000 vs - processSlashings	33.791 us/op	3.5590 ms/op	0.01
epoch - 250000 vs - processFinalUpdates	39.957 ms/op	39.089 ms/op	1.02
epoch - 250000 vs - prepareEpochProcessState	543.03 ms/op	1.0793 s/op	0.50
getAttestationDeltas - 250000 vs	110.40 ms/op	165.88 ms/op	0.67
processSlots - 250000 vs - 32 empty slots	3.5484 s/op	1.9698 s/op	1.80
shuffle list - 512 els	144.20 us/op	1.4239 ms/op	0.10
shuffle list - 16384 els	1.8196 ms/op	15.937 ms/op	0.11
shuffle list - 250000 els	24.698 ms/op	225.32 ms/op	0.11
aggregationBits - 2048 els - readonlyValues	235.97 us/op	273.96 us/op	0.86
aggregationBits - 2048 els - zipIndexesInBitList	43.099 us/op	56.178 us/op	0.77
ssz.Root.equals	1.2270 us/op	1.2810 us/op	0.96
BLS verify - blst-native	1.9296 ms/op
BLS verifyMultipleSignatures 3 - blst-native	3.9775 ms/op
BLS verifyMultipleSignatures 8 - blst-native	8.5658 ms/op
BLS verifyMultipleSignatures 32 - blst-native	33.578 ms/op
BLS aggregatePubkeys 32 - blst-native	42.014 us/op
BLS aggregatePubkeys 128 - blst-native	161.04 us/op
CheckpointStateCache - add get delete	34.588 us/op

by benchmarkbot/action