fix(charging): limit charge steps and show planning failures

[frahlg]

Problem and result

A charger with an 11,000 W limit and an 11,100 W allowed step passed config validation, then stopped every planning attempt. Plan stayed on “Preparing” and hid the cause.

FTW now reduces steps above the charger limit to that limit and removes duplicates. Loading, saving and planning all apply this rule. Settings saves the corrected values, puts them back in the field and shows a warning. Plan keeps working and shows the warning. Other planning failures now show their actual cause; a successful retry clears the error.

Scope and safety

The correction only lowers charging power. Core still validates the resulting plan and commands, and stale meter data still stops dispatch. Failed replans keep old diagnostics but do not resume dispatch from the old plan. An older attempt cannot overwrite the latest error or warning.

Based on current master, d570ab7. Open PR #1177 also touches web/plan.js: its Plan polling changes already entered master in #1225 (e5f281f). This patch keeps that code and its tests intact. #1003 and #1051 change other config fields; #1337 and #1595 add other API handlers.

Verification

• GitHub CI passed on a4717ffe: all seven workflows, including Go, web, full-stack tests and both Linux builds.
• make verify passed: Go tests, vet, build, protocol boundary, install/release fixtures and bundled Energyplan checks.
• npm test: 719 tests passed.
• Focused Go race checks passed for planning error recovery, stale attempts, charge-step correction and replan cancellation.
• Tests cover the reported step list, YAML and SQLite persistence, API save feedback, planner inputs and power limits, error recovery, and corrected values on the next Settings save.
• Browser check at 390 × 844 and 1280 × 900: Settings returns 11,000 W after entering 11,100 W and shows the warning; an accepted plan shows its warning; a failed plan shows its error. This used the actual UI with synthetic API data. No physical charger or customer box was used.

The repository owner requested merge and a beta release on 2026-10-13 with green CI. This uses the owner fast-track allowed by AGENTS.md. Human browser review remains deferred; the local preview uses test data.

Checklist

• The change follows VISION.md and one selected scope.
• I checked shared files and kept the merged polling work.
• Relevant checks cover the changed behaviour and failure paths.
• A human reviewed changed UI in a browser.
• A Changeset is included.
• Every commit has a DCO sign-off.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-13T05:31:41.491424Z	a4717ff	Draft marked ready

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" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.