Improve adaptive forecasts and show allowance restorations

[TheBeems]

Summary

• apply saved adaptive weekly history to dashboard projections and prevent paused intervals from being learned retroactively
• detect weekly allowance restorations, including early window resets, while excluding scheduled rollovers
• show detected restorations as amber chart markers, a latest-event summary, and an active-window Details history
• persist window identity metadata compatibly with existing local weekly history
• update user and developer documentation

User impact

Weekly usage changes that restore allowance are now explicit instead of appearing only as an unexplained line break. Forecasts restart from the latest restoration and continue to treat current pace as the primary signal.

Validation

• x64: 101 tests passed; Debug self-contained build succeeded with zero warnings
• ARM64: 101 tests passed; Debug self-contained build succeeded with zero warnings
• ARM64 development registration and manifest, COM, and AppExtension integration preflight passed
• git diff --cached --check passed before commit

Remaining manual check

Reload Command Palette Extension and visually verify restoration marker contrast and text wrapping in narrow and wide layouts, light and dark themes, and high contrast.

[chatgpt-codex-connector]

💡 Codex Review

https://github.com/TheBeems/CodexUsageDock/blob/23816c812b48431a0f1818766eb8f532d4d9e56c/CodexUsageService.cs#L202
Preserve baseline requests until a weekly sample exists

When adaptive forecasting is re-enabled while the current snapshot has no weekly window (for example, data is temporarily unavailable or only a five-hour limit is returned), AdvanceBaseline returns without updating LastSample, but this line still clears _adaptiveWeeklyForecastNeedsBaseline. The next valid weekly refresh then falls into Record() and replays all stored samples after the old baseline, so measurements collected while the forecast was paused can be learned despite the setting’s documented behavior; keep the flag set unless a baseline was actually advanced.

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".