Codeblock Regression

Goal

Reproduce the reported code block regression in the running docs app, identify the root cause, and fix it with fresh verification.

User reports a codeblock regression and asks whether it was verified in browser-use.
Current browser surface is http://localhost:3001/.
I did not specifically verify code block behavior in browser before the report.

docs/solutions/logic-errors/2026-03-23-code-block-tab-should-indent-every-selected-line.md
docs/solutions/logic-errors/2026-03-26-code-block-language-change-must-trigger-redecorate.md
docs/solutions/logic-errors/2026-03-27-code-block-format-must-rebuild-code-lines.md
docs/solutions/logic-errors/2026-03-28-link-validation-must-not-treat-double-slash-as-internal-path.md
docs/solutions/logic-errors/2026-04-03-editor-key-protocols-must-cover-expanded-selection-and-repeated-escalation.md

Code block already has multiple recent regressions around formatting, redecorate, paste handling, and key protocols. That makes it a likely regression-prone surface, not a one-off complaint.
No current learning directly names the user's exact regression yet.
localhost:3001 is not a trustworthy repro surface right now. The active docs dev server is hanging behind Turbopack cache corruption and missing .sst files under apps/www/.next/dev/cache/turbopack.
A clean docs instance on localhost:3002 reproduced the real browser-facing regression on /docs/code-block: the route fell into the "This page couldn’t load" error state with PlateError: [CODE_HIGHLIGHT] Error: Could not highlight with Highlight.js.
Root cause: setCodeBlockToDecorations tried to recover from registered language highlight failures, but called editor.api.debug.error(...) inside the catch. In dev, debug.error throws, so the intended plaintext fallback never ran.
After the patch, /docs/code-block loads successfully in browser. The page now logs a warning and falls back to plaintext for the Python sample instead of crashing the route.