fix(tui): use current model for context-limit display after switch

[chethanuk]

Issue for this PR

Addresses upstream #35072. Opened against my own fork first as a staging/review step — not raised upstream yet.

Type of change

• Bug fix
• New feature
• Refactor / code improvement
• Documentation

What does this PR do?

The context-limit percentage derived its denominator from the last assistant message's model. After a mid-session model switch it kept showing the previous model's limit until a new assistant message arrived — e.g. switching from a 128k model to a 32k model still displayed the 128k-based percentage.

The fix uses the currently-selected model (local.model.current()) as the denominator, falling back to the last assistant message's model when nothing is selected yet:

• packages/tui/src/component/prompt/index.tsx — composer footer usage memo.
• packages/tui/src/feature-plugins/sidebar/context.tsx — the sidebar meter named in the issue. The sidebar can display a subagent session, where the global model picker is unrelated to the subagent's own model, so it uses current() only for top-level sessions (via a session()?.parentID guard) and otherwise keeps the session's own last-message model.

Token count still comes from the last message (real occupancy); only the limit denominator changed. subagent-footer.tsx is intentionally left unchanged — a completed subagent's model is fixed, so its last-message model is already the correct denominator there.

How did you verify your code works?

• bun run typecheck (tsgo) in packages/tui passes clean.
• Traced the switch flow: local.model.current() resolves to the just-selected model, so the denominator updates on the next render instead of waiting for an assistant reply. The ?? last-message fallback preserves the pre-existing behavior when no model is selected yet, and both call sites read current() inside a createMemo, so reactive tracking is preserved.

Screenshots / recordings

TUI-only change; no screenshot.

Checklist

• I have tested my changes locally
• I have not included unrelated changes in this PR

[github-actions]

Thanks for your contribution!

This PR doesn't have a linked issue. All PRs must reference an existing issue.

Please:

1. Open an issue describing the bug/feature (if one doesn't exist)
2. Add Fixes #<number> or Closes #<number> to this PR description

See CONTRIBUTING.md for details.

[github-actions]

The following comment was made by an LLM, it may be inaccurate:

POTENTIAL DUPLICATE FOUND:

• PR fix(tui): use current model for sidebar context limit after model switch #35287: fix(tui): use current model for sidebar context limit after model switch
  • fix(tui): use current model for sidebar context limit after model switch #35287

Why it's related: PR #35287 appears to address the same issue (issue #35072) with nearly identical scope: fixing the context-limit display in the TUI sidebar to use the current model after a model switch. The title is extremely similar to PR #35542, and both target the same files and fix the same root cause.

Please check PR #35287 to determine if it's already merged or if PR #35542 is a duplicate submission.