Add Intermediate 'Steps Only' Thinking Display Mode for Codex CLI

What feature would you like to see?

I'd like to propose adding a new display mode for the Codex CLI reasoning summary format that strikes a balance between the two existing modes: 'none' and 'experimental'.

• In 'none' mode, there is no feedback on the thinking process, so users can't see what steps are being taken.
• In 'experimental' mode, the output is very verbose, showing all internal contents of the thinking process, which can be overwhelming.

A better approach would be to borrow the display style from the VS Code extension: for each thinking step, only display its title or block header, rather than all content. This would make the thinking process transparent and understandable, but concise and less cluttered than 'experimental' mode.

Additional information

This 'steps only' mode would help users track what the model is doing without overwhelming them with details—essentially listing the titles of each thinking step as they are executed. It would bring the Codex CLI UX closer to that of the VS Code extension, promoting a unified and user-friendly experience.