Keep startup prompts interactive

[nightt5879]

Summary

• Treat top-level prompt input as an interactive startup prompt instead of running it through one-shot mode.
• Add explicit startup input modes so codewhale pr still pre-fills editable text while codewhale --prompt auto-submits the initial request.
• Keep startup prompt dispatch robust for multibyte text and onboarding-gated startup flows.

Testing

• cargo fmt --all -- --check
• cargo clippy --workspace --all-targets --all-features
  • Passed with existing warnings outside this patch.
• cargo test -p codewhale-cli --lib --all-features
• cargo test -p codewhale-cli --lib --all-features top_level
• cargo test -p codewhale-tui --bin codewhale-tui --all-features prompt_flag_starts_interactive_submit_input
• cargo test -p codewhale-tui --bin codewhale-tui --all-features initial_input_
• cargo test -p codewhale-tui --bin codewhale-tui --all-features startup_prompt_waits_for_onboarding_then_dispatches
• cargo test --workspace --all-features
  • Passed by PR CI on ubuntu-latest, macos-latest, and windows-latest.
• GitHub Actions CI
  • Lint, Version drift, Test (ubuntu-latest), Test (macos-latest), Test (windows-latest), npm wrapper smoke, GitGuardian, and Greptile Review all passed.

Checklist

• Updated docs or comments as needed
• Added or updated tests where relevant
• Verified TUI behavior manually if UI changes
  • Covered by direct command-line equivalent CLI/TUI startup tests in this Windows session: prompt_flag_starts_interactive_submit_input, initial_input_, and startup_prompt_waits_for_onboarding_then_dispatches.

Closes #2370

[gemini-code-assist]

Code Review

This pull request changes the top-level prompt flag (-p / --prompt) behavior to start an interactive TUI session and automatically submit the initial prompt, rather than running in a non-interactive one-shot mode. Non-interactive runs are now directed to use the exec command. To support this, an InitialInput enum is introduced with Prefill and Submit variants. The review feedback correctly identifies a potential bug where text.len() is used to set the cursor position instead of text.chars().count(), which could lead to incorrect cursor placement or panics when handling multi-byte UTF-8 characters.

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 7b0562a10f

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".

[nightt5879]

Updated the PR description after follow-up validation.

Review feedback addressed in 77bb447a:

• Startup prompt cursor now uses character counts, so multibyte input is handled correctly.
• Deferred startup prompts now remain pending through onboarding and auto-submit once onboarding completes, avoiding the stale setup message path.

Re-ran direct command-line equivalent CLI/TUI behavior tests on this Windows checkout:

• cargo test -p codewhale-tui --bin codewhale-tui --all-features prompt_flag_starts_interactive_submit_input
• cargo test -p codewhale-tui --bin codewhale-tui --all-features startup_prompt_waits_for_onboarding_then_dispatches
• cargo test -p codewhale-tui --bin codewhale-tui --all-features initial_input_
• cargo test -p codewhale-cli --lib --all-features top_level

Full workspace validation is green in PR CI on Ubuntu, macOS, and Windows.