feat: syntax highlighting for tool outputs and LLM responses

[KUSH42]

feat: syntax highlighting for tool outputs and LLM responses

⚠️ Stacked on feat/rich-diff-renderer — merge that first. All commits above the tip of that branch are new here.

Wires the SyntaxHighlighter / LanguageDetector from the diff-renderer PR into tool result display and LLM response rendering.

Changes

execute_code preview

• Highlighted Python block printed after successful execution
• Success-gated via _result_succeeded — no highlight shown after a failed run
• Cute-msg drops the inline code snippet when highlight is active to avoid duplication

read_file preview

• ┊ 📄 filename.py header + syntax-highlighted content
• Language detected from file extension only (no content guessing)

terminal preview

• Verb-based language detection from the command
• _FILE_EXEC_COMMANDS blocklist (node, python3, bash, ruby, …) — runtime stdout is never mistaken for source code

LLM response rendering

• format_response() for complete responses: replaces fenced code blocks with ANSI-highlighted versions
• StreamingCodeBlockHighlighter state machine for streaming: buffers lines inside fences, flushes the highlighted block on the closing fence, passes plain text through immediately. Language hint used when present, content-detection fallback otherwise.

Plumbing

• Verbosity gate: all previews early-return when tool_progress_mode == "off"
• Config key code_highlight (display.code_highlight in config.yaml) + /code-highlight toggle
• set_code_highlight_active() in display.py keeps it decoupled from CLI state

Bug fixes

• Fence delimiters in Panel output — format_response preserved ``` fence markers around highlighted code blocks ("so the Panel still looks like a code block"). In practice the ANSI-highlighted block reads cleanly without them, and keeping the fences caused raw backtick lines to appear in the rendered response view. Removed; the regex replacement now returns the highlighted code directly.

• LLM response markdown gated on /code-highlight toggle — apply_block_line / apply_inline_markdown in _emit_stream_text and _flush_stream were guarded by _display._code_highlight_active, the same flag that controls tool-output syntax highlighting. Toggling /code-highlight off silently suppressed bold, italic, headings, blockquotes and all inline markdown in LLM responses. Fixed by removing the inner guard — _RICH_RESPONSE (import guard) is the correct gate for response rendering.

• CommonMark backslash escapes visible in output — LLM responses containing escaped punctuation (e.g. \] in task-list syntax [ \]) showed the raw backslash. Added a step-7 pass in apply_inline_markdown that strips all CommonMark backslash escapes (\] → ], \* → *, etc.) after code spans are restored, so protected spans are never affected.

Tests

135 passing (tests/test_display.py + tests/test_rich_output.py).

[KUSH42]

Superseded by NousResearch#4471