feat(tui): render next prompt suggestions [3 of 3]

[fcoury-oai]

Why

After the app-server can produce a next-prompt suggestion, the TUI needs to request it at the right moments and show it as lightweight composer ghost text without submitting on behalf of the user.

What changed

• Request next-prompt suggestions after completed primary-thread turns and after resume/fork/new-thread attachment.
• Ignore stale results and cancel pending suggestion work when the user types, pastes, rewinds, or changes thread state.
• Preserve the last visible suggestion when the user clears a draft, while clearing stale suggestions on rollback/backtrack.
• Render suggestions as empty-composer ghost text and accept them with Tab without immediately submitting.
• Add focused TUI coverage and the empty-composer snapshot.

Testing

• RUST_MIN_STACK=8388608 CARGO_TARGET_DIR=/private/tmp/codex-next-suggestion-stack-target cargo test -p codex-tui next_prompt_suggestion
• cargo insta pending-snapshots --manifest-path tui/Cargo.toml
• CARGO_TARGET_DIR=/private/tmp/codex-next-suggestion-stack-target ./tools/argument-comment-lint/run.py -p codex-core -p codex-app-server-protocol -p codex-app-server -p codex-tui

Manual smoke retained from the original combined PR: a fresh temporary-home TUI session produced the expected ghost-text suggestion after a completed turn; resumed thread attachment invoked thread/suggestNextPrompt; incomplete active tool flow suppressed suggestions instead of showing stale text.

Stack

1. #24126 next-prompt suggestion engine
2. #24127 next-prompt app-server RPC
3. feat(tui): render next prompt suggestions [3 of 3] #23976 next-prompt TUI (this PR)

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 6c55c12f7e

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".