Improve chat debugging and HTML reporting

[aniket-s-kulkarni]

Summary

This PR expands the dremio chat debugging workflow so chat runs can be inspected in the terminal or exported as a standalone HTML report.

What Changed

Richer transcript output

• Added --show-tool-details to one-shot dremio chat -m ... output.
• Added --show-tool-details to dremio chat history.
• Transcript output now shows:
  • tool start and finish timestamps
  • per-tool duration
  • full tool result content instead of only compact status lines

Better chat history handling

• Fixed chat history rendering to behave consistently for conversation message envelopes.
• Added support for fetching the full paginated message history for HTML export.

Terminal Gantt improvements

• Added transparent-background Gantt rendering that adapts foreground colors for light and dark terminal themes.
• Kept --think-time support for synthetic gaps between steps.

Standalone HTML chat reports

• Added dremio chat html.
• Supports:
  • one or more positional conversation IDs
  • repeated --dump-file inputs
  • -o / --output
  • optional --think-time
• The HTML report is self-contained and embeds all data in the page.
• The report includes:
  • an overview page
  • aggregate stats across all selected conversations
  • mean, median, standard deviation, min, max, and totals for runtime metrics
  • navigation from the overview page to each conversation’s Gantt chart
  • conversation summary and result details
  • selected tool span payload and tool result details

Packaging fix

• Added a fallback Hatch version for source trees without usable git metadata.
• This allows uv sync / editable installs to work on machines where the checkout is missing full .git metadata, instead of failing during version resolution.

Why

• Chat runs were difficult to debug from the CLI because tool activity was summarized too aggressively.
• The original HTML export work needed an overview page and richer details to be useful across multiple conversations.
• Source installs were fragile on fresh machines or copied source trees because hatch-vcs required VCS metadata.

Impact

• Developers can inspect AI chat runs directly from the CLI with timestamps and tool outputs.
• Multi-conversation debugging is easier with a portable standalone HTML report.
• Local setup from a non-git source tree is more resilient.

Validation

• uv run pytest tests/test_cli.py tests/test_chat_gantt_html.py
• Verified uv sync succeeds from a git-less copy of the repo after adding the fallback version

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 82b87e33c3

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".