feat(web): response-viewer transcript fallback + code-block rendering

[TeigenZhang]

Summary

Improves the response viewer when the JSONL transcript is unavailable (it has to fall back to the raw terminal buffer) and hardens code/diagram rendering.

Changes

• app.js _cleanTerminalBuffer(): strip ANSI escapes + Claude CLI chrome (status bar, spinner, progress bar, prompt glyphs) so the fallback renders clean conversational text
• app.js _preprocessAsciiArt(): wrap box-drawing/block-element diagrams in fenced code blocks (narrow trigger excludes arrows/geometric shapes common in prose) so marked.js preserves whitespace
• styles.css: extend .rv-text rules to .response-viewer-body so fallback-rendered content gets the same typography, code-block, and table styling

Notes

Built on top of the existing response-viewer (copy button / wrap toggle already upstream); this fills the transcript-missing gap.

Test plan

• tsc --noEmit passes (no .ts touched)
• Open a response with no JSONL transcript, confirm clean render
• Render a box-drawing diagram, confirm whitespace preserved

[Ark0N]

Thank you as always for contributing, @TeigenZhang! 🙏

The response-viewer CSS overhaul is lovely — proportional prose font with monospace kept for code, the heading/blockquote/code styling, the readable max-width, and the smoother slide-in animation all make the viewer much nicer to read. The _renderMarkdown null-safety fix (text → src = text || '') is a good catch too.

One heads-up from review: this branch forked before realizing that _cleanTerminalBuffer and _preprocessAsciiArt already landed on master back in #75, so it had re-added byte-identical copies of both. In a JS class body the later definition wins, so the duplicates were inert dead code. I pushed a small cleanup commit (bdeab49) removing them, leaving just the genuinely-new CSS + null-safety work.

Verified after cleanup: each method now defined exactly once, tsc/lint/format/build all clean, node --check clean, and a test-merge confirms #100/#101 are preserved. Merging now.