Reduce pull request performance checks

[SandroMaglione]

Summary

• classify pull request changes so type performance, runtime performance, memory measurement, and examples run only when relevant
• retain all 13 Effect Machine runtime scenarios and all 7 Effect Machine memory profiles while removing XState and Effect-runtime comparison baselines
• keep five interleaved base/head processes and fail only large regressions beyond fixed and variability-adjusted thresholds
• preserve stable required gate names when measurements or examples are skipped
• stop duplicating CI on main and limit release runs to publishable changes

Changeset

• Added for the package-metadata and benchmark dependency changes

Validation

• pnpm check
• pnpm perf:types
• pnpm perf:runtime
• clean-install pnpm perf:runtime -- --quick
• workflow YAML parsing and focused classifier/regression tests

The focused full runtime suite completed locally in about 14 seconds for one
benchmark process while retaining every Effect Machine memory profile.

GitHub validation completed successfully:

• CI: 1m28s, including all three examples selected by this infrastructure change
• type performance: 1m34s
• runtime and memory performance: 4m16s total, with 4m1s spent measuring five interleaved base/head processes
• trusted type and runtime sticky-comment workflows: passed

The previous representative runtime workflow took 10m13s, including a 9m35s
measurement step. After this PR merges, runtime-performance can be added to
the required status checks using the stable gate introduced here.

[github-actions]

Type performance

Measured with TypeScript 6.0.3 and skipLibCheck=true.

Scenario	Base	PR	Difference
Effect only	55	55	0 (0.0%)
Import effect-machine	55	55	0 (0.0%)
Machine.defineStates (3 states)	2,965	2,965	0 (0.0%)
Machine.make (3 states, 2 events)	8,715	8,715	0 (0.0%)
machine.handle (3 states, 2 transitions)	23,492	23,492	0 (0.0%)
machine.handle (depth 24)	183,851	183,851	0 (0.0%)
machine.handle (wide depth 16)	214,189	214,189	0 (0.0%)
machine.handle (parallel/history/choice)	132,667	132,667	0 (0.0%)
machine.handle (4 successive calls)	128,674	128,674	0 (0.0%)
machine exact input/output/error/services	110,584	110,584	0 (0.0%)
execution adapter readiness	127,854	127,854	0 (0.0%)

Marginal instantiations are measured against the matching setup without that API call:

Scenario	Base	PR	Difference
Import effect-machine	0	0	0
Machine.defineStates (3 states)	2,910	2,910	0 (0.0%)
Machine.make (3 states, 2 events)	5,742	5,742	0 (0.0%)
machine.handle (3 states, 2 transitions)	14,777	14,777	0 (0.0%)
machine.handle (depth 24)	171,577	171,577	0 (0.0%)
machine.handle (wide depth 16)	202,224	202,224	0 (0.0%)
machine.handle (parallel/history/choice)	114,331	114,331	0 (0.0%)
machine.handle (4 successive calls)	113,703	113,703	0 (0.0%)
machine exact input/output/error/services	100,670	100,670	0 (0.0%)
execution adapter readiness	102,092	102,092	0 (0.0%)

Check times (informational)

Scenario	Base	PR
Effect only	0.03s	0.03s
Import effect-machine	0.04s	0.03s
Machine.defineStates (3 states)	0.11s	0.10s
Machine.make (3 states, 2 events)	0.16s	0.16s
machine.handle (3 states, 2 transitions)	0.25s	0.23s
machine.handle (depth 24)	0.70s	0.69s
machine.handle (wide depth 16)	0.76s	0.76s
machine.handle (parallel/history/choice)	0.61s	0.63s
machine.handle (4 successive calls)	0.57s	0.58s
machine exact input/output/error/services	0.51s	0.51s
execution adapter readiness	0.61s	0.57s

Type instantiations are the comparison metric. Check time varies with runner load and is informational only.

[github-actions]

Runtime performance

Median of 5 independent benchmark processes on AMD EPYC 9V74 80-Core Processor with Node v24.18.0.

Pull request baseline

Scenario	Effect Machine
Plan counter transitions	129,627 transitions/s
Drain burst with terminal fence	518,323 increments/s
Drain burst with a change observer	478,528 increments/s
Lookup and send to one child	425,536 increments/s
Start and stop a machine	166,141 machines/s
Start and stop a parent with one child	44,045 families/s
Plan transitions through a compound state	115,946 transitions/s
Plan transitions through parallel regions	91,843 transitions/s
Drain burst through a compound state	447,429 events/s
Drain burst through two parallel regions	413,505 events/s
Drain a compound-state burst with a change observer	415,080 events/s

Effect runtime reference points

Scenario	Effect Machine
Start and stop a raw generic process	27,215 processes/s
Start and stop a raw compiled process	70,817 processes/s

Memory profile	Effect Machine
Idle machine	1.5 KiB
Raw generic managed process	12.7 KiB
Raw compiled process	2.8 KiB
Two independent idle machines	3.0 KiB
Idle parent with one child	4.8 KiB
Parent with observed child registry	9.0 KiB
Parent with observed invoked child snapshots	5.0 KiB

Effect Machine change from base

Metric	Base	Base variability	PR	PR variability	Difference
Plan counter transitions	130,520 transitions/s	0.4% MAD	129,627 transitions/s	1.2% MAD	-0.7%
Drain burst with terminal fence	525,392 increments/s	0.3% MAD	518,323 increments/s	1.7% MAD	-1.3%
Drain burst with a change observer	482,627 increments/s	0.2% MAD	478,528 increments/s	1.4% MAD	-0.8%
Lookup and send to one child	429,137 increments/s	0.2% MAD	425,536 increments/s	2.0% MAD	-0.8%
Start and stop a machine	167,842 machines/s	0.3% MAD	166,141 machines/s	0.1% MAD	-1.0%
Start and stop a parent with one child	45,080 families/s	0.5% MAD	44,045 families/s	0.4% MAD	-2.3%
Plan transitions through a compound state	117,094 transitions/s	0.0% MAD	115,946 transitions/s	1.5% MAD	-1.0%
Plan transitions through parallel regions	92,386 transitions/s	0.1% MAD	91,843 transitions/s	2.2% MAD	-0.6%
Drain burst through a compound state	447,968 events/s	0.8% MAD	447,429 events/s	1.9% MAD	-0.1%
Drain burst through two parallel regions	412,555 events/s	0.2% MAD	413,505 events/s	0.8% MAD	+0.2%
Drain a compound-state burst with a change observer	414,212 events/s	1.0% MAD	415,080 events/s	2.3% MAD	+0.2%
Idle machine heap per unit	1.5 KiB	0.1% MAD	1.5 KiB	0.1% MAD	-0.2%
Raw generic managed process heap per unit	12.7 KiB	0.0% MAD	12.7 KiB	0.0% MAD	+0.0%
Raw compiled process heap per unit	2.8 KiB	0.1% MAD	2.8 KiB	0.2% MAD	-0.0%
Two independent idle machines heap per unit	3.0 KiB	0.0% MAD	3.0 KiB	0.1% MAD	+0.1%
Idle parent with one child heap per unit	4.8 KiB	0.0% MAD	4.8 KiB	0.0% MAD	+0.0%
Parent with observed child registry heap per unit	9.0 KiB	0.0% MAD	9.0 KiB	0.0% MAD	+0.0%
Parent with observed invoked child snapshots heap per unit	5.0 KiB	0.1% MAD	5.0 KiB	0.1% MAD	+0.2%

Effect runtime reference change from base

Metric	Base	Base variability	PR	PR variability	Difference
Start and stop a raw generic process	27,259 processes/s	1.9% MAD	27,215 processes/s	1.4% MAD	-0.2%
Start and stop a raw compiled process	71,018 processes/s	0.3% MAD	70,817 processes/s	0.6% MAD	-0.3%

Versions and interpretation

• Effect Machine: 0.6.0

Higher throughput is better; lower heap is better. Variability is the median absolute deviation across independent processes, relative to their median. Runtime measurements on shared GitHub-hosted hardware remain informational, so small differences should be confirmed across multiple workflow runs.