refactor(metrics): Complete coherent energy sampling

[bigtomcat6]

Summary

• Replace split Energy Impact sampling/publication state with one coherent three-second sampler observation.
• Move process reads, ownership, baselines, publication statistics, smoothing, ranking, and sorting behind an actor-owned lease.
• Reconcile the safety and coverage regressions from PR refactor(metrics): Isolate visible sampling sessions #113 under the coherent lease architecture.
• Add an inert-by-default, repeatable five-process native cost gate.
• Supersede the unmerged local Part 4 implementation at feat/energy-impact-coherent-sampling; that retained branch remains unchanged as historical evidence and is not the PR head.
• Integrated series:
  • 85ff0092af25544791d47c42743140f5a3f2e934 refactor(energy): extract publication state
  • a31bd63d0b68886cd14872a233bc9c84a4962999 refactor(energy): add coherent sampler sessions
  • 88041b5fb0dd488e42b32ac816564dde4d47a0bd refactor(energy): control visible observation lifecycle
  • abf4ecb884005fae2b28f4fca2e25f442bcbbb63 test(energy): add repeatable native gate

How to test

• swift test --filter EnergyImpact
  • Exact result: 126 total / 125 passed / 1 expected native-default skip / 0 failed / 0 unexpected failures.
• swift test
  • Exact result: 1,170 total / 1,169 passed / 1 expected native-default skip / 0 failed / 0 unexpected failures.
• xcodegen generate --quiet
  • PASS: regenerated MacActivity.xcodeproj/project.pbxproj was unchanged.
• xcodebuild test -project MacActivity.xcodeproj -scheme MacActivity -destination 'platform=macOS' CODE_SIGNING_ALLOWED=NO
  • ** TEST SUCCEEDED **; xcresult: 1,170 total / 1,169 passed / 1 skipped / 0 failed / 0 expected failures.
• Static superseded-API and MainActor-boundary scans from the Part 4 integration gate returned zero matches: no Energy topApps call, no superseded cadence/session/cache symbols, and no App/provider ownership of low-level process or smoothing/ranking state.
• Five-process native gate, authoritative attempt 3 (each run had 21 observations and passed the p95, CPU, and wall-duration thresholds):

Run	Observations	Pre-run apps	Snapshot processes	p50 ms	p95 ms	CPU %	Wall s
1	21	7	398	13.427	15.319	0.438875	60.092
2	21	7	397	13.128	15.377	0.452884	60.099
3	21	7	397	13.811	15.028	0.456702	60.043
4	21	7	397	12.498	15.778	0.413864	60.056
5	21	7	400	13.572	15.199	0.449276	60.116

• Aggregate result: median CPU 0.449276%; maximum CPU 0.456702%; maximum p95 15.778 ms; minimum wall duration 60.043 s; valid native runs 5/5; ENERGY_NATIVE_GATE PASS.
• Attempts 1 and 2 are preserved only as non-authoritative failed/incomplete execution-environment attempts and were not used for this PASS.
• Manual visible lifecycle limitation: the SHA-bound Debug app built and launched, but Computer Use returned timeoutReached for the app name and exact app path; the bundle ID was ambiguous across builds. No accessibility tree or screenshot was returned, so no visible scenario is claimed as passed:
  1. Immediate honest collecting/status rows without false zero: UNVERIFIED — the menu-bar UI was not observable.
  2. Next valid value opportunity after approximately three seconds: UNVERIFIED — no UI state or timestamped row transition was observable.
  3. Rapid Energy → another tab → Energy suppresses an old row from the prior run: UNVERIFIED — the required tab interactions were unavailable.
  4. Return after at least six seconds hidden without queued/burst publication: UNVERIFIED — the visibility transition was unavailable; internal read stoppage is also manually unverified, with deterministic lifecycle tests providing internal-contract evidence.
  5. Sixty-second visible observation with no burst catch-up: UNVERIFIED — no visible observation could be started.

Release impact

• Type: Internal
• Release note: None

Checklist

• I have seen this code, I have run this code, and I take responsibility for this code.
• PR title follows type(scope): Summary.
• Tests or a clear manual verification path are included.
• User-facing changes include release-note wording or Release impact: None.

[bigtomcat6]

@codex review

[codecov]

Codecov Report

❌ Patch coverage is 92.66504% with 30 lines in your changes missing coverage. Please review.

Files with missing lines	Patch %	Lines
...yCore/Metrics/Providers/EnergyImpactProvider.swift	41.86%	25 Missing ⚠️
...rces/MacActivityApp/Models/EnergyImpactModel.swift	94.23%	3 Missing ⚠️
...tyCore/Metrics/Providers/EnergyImpactSampler.swift	98.71%	2 Missing ⚠️

📢 Thoughts on this report? Let us know!

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: abf4ecb884

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[bigtomcat6]

reviewed