fix(support): report control plan identity

[frahlg]

What changed

• carry the MPC decision ID into the control slot used by each battery tick
• carry the same ID through the supported legacy dispatch rollback
• reuse one accepted plan revision for every plan-dependent check in a tick
• include the latest plan ID, control slot ID, and match state in the existing help report
• flag a plan/control ID mismatch as a support finding

Why

The old help file could show the current plan and live values, but it could not prove which plan revision the control tick used. That made plan drift hard to rule in or out in cases such as passive arbitrage exporting while a later charge slot remained in the plan.

This replaces the former broad diagnostics draft. It adds no endpoint, log archive, telemetry store, command audit, or new privacy surface.

Checks

• focused plan identity tests x100
• focused race tests x20
• full race tests for affected Go packages
• make verify

Release order

This is based on master after #922. The later explicit-zero and slew fixes (#887/#928) overlap control files and should rebase after this small support change.

[frahlg]

@codex review

@miravoss26 please review exact head 03a7a0cf5bd492ea3afb4bf2400c9897ca6ff905, based on live master 4b7ad49ebae8a59563ba449b5cf30947fc5322ce. This replaces the prior broad draft with one small help-report fix: it reports the plan ID used by the last battery control tick and flags plan/control mismatches. No new endpoint, archive, telemetry store, command audit, or privacy surface. Focused tests passed x100, focused race x20, affected-package race and make verify are green. The tree is byte-identical to the tested 5b96f188; only the commit ID changed to trigger master PR checks after retargeting.

[frahlg]

@codex review

Review refresh for exact head 03a7a0cf5bd492ea3afb4bf2400c9897ca6ff905. The tree is unchanged from the tested patch; only the commit ID was renewed after retargeting so GitHub runs the master PR workflows.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 5b96f18848

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".

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 03a7a0cf5b

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".

[frahlg]

@codex review

@miravoss26 review refresh for exact head fe776a497348b73d6cf34c9ff7b203408a90af66. The P2 now carries the accepted decision ID through the existing legacy PlanTarget callback, so planner.legacy_dispatch and deprecated use_energy_dispatch: false report the plan used by the battery tick too. Added control and MPC identity tests. Focused identity tests passed x100, focused race x20, full affected-package race and make verify are green.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: fe776a4973

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".

[frahlg]

@codex review

@miravoss26 final review refresh for exact head 07f3d1ee6cfdb59752802697a9cffa4401f9f735. The second P2 is fixed: each battery control tick caches the first slot directive and legacy target, then every plan-dependent guard uses the same selected source and decision ID. Tests require exactly one directive or legacy-target read per tick, including same-slot replans and stale-plan transitions. Focused x100, focused race x20, full control race and make verify are green.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 07f3d1ee6c

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".

[frahlg]

@codex review

@miravoss26 final refresh for exact head 658f90473ed4692e0adf88a93029bc797b5da9b0, now based on live master abb850d6289c032dff68437690e1b5f5ed753a29 after #798. All four #927 patches match the pre-rebase versions exactly by range-diff. The latest P2 is fixed by using the cached energy directive as the legacy sign guard only when its decision ID matches the legacy target ID; mismatched generations fall back to the legacy intent. Focused x100, race x20, full affected-package race and make verify are green.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 658f90473e

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".

[frahlg]

@codex review

@miravoss26 review refresh for exact head 90c88185460a1a9c15766b3bc3d5eb6b1d5fd549, based on live master abb850d6289c032dff68437690e1b5f5ed753a29. The exact-head P2 is fixed by capturing one time at control-tick start and using it for both the slot directive and legacy plan target. The regression was red on 658f904 across a slot boundary, then passed x100; the focused race passed x100, the full affected-package race and make verify are green. No new endpoint, archive, telemetry store, command audit, or privacy surface.

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Keep them coming!

Reviewed commit: 90c8818546

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".

[frahlg]

@codex review

@miravoss26 please review exact head 01cbe32afa2195ad8e5750752f7cdce4e09c29fc, now based on live master c3d37017fc7a926d90e1956ae5f6092fc01aacad. The four later patches are range-diff identical. The first patch keeps #929 runtime and release identity code intact and adds only the MPC-to-control DecisionID adapter; that is the sole conflict resolution. Boundary, identity, sign, API, and MPC tests passed x100 and race x100. The full affected-package race and make verify pass after syncing the ignored driver snapshot to the new #930 pin. All five older review threads now contain exact evidence and are resolved. No merge is requested.

[chatgpt-codex-connector]

Codex Review: Didn't find any major issues. Keep them coming!

Reviewed commit: 01cbe32afa

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".