test(terminal-report-filter): cover synthetic + replay report validators

[dormouse-bot]

Surfaced by the nightly code-quality survey.

terminal-report-filter.ts had test coverage only for stripMouseReportsFromInput. Its two most complex exports — inputIsSyntheticTerminalReport and inputIsReplayTerminalReport — had none anywhere in the repo, despite being consumed by the replay/lifecycle path to decide whether inbound data is a synthetic terminal report that should be filtered rather than shown to the user.

Both are tokenizer/validators built from concatenated CSI/SS3/OSC (and, for replay, DCS/focus) regexes with an all-chunks-must-validate rule. A regression in the token set or the ^(?:…)$ anchoring — e.g. loosening the . fallback — could let real user input be misclassified as a synthetic report and silently dropped, with nothing to catch it.

This PR adds tests pinning the current behavior:

• Whole-string-is-reports acceptance (DSR, CPR, SS3, OSC; multiple concatenated reports).
• Rejection of empty input and of any input containing plain-text bytes (the every-chunk-must-validate contract, including a report with a trailing stray char).
• For the replay validator specifically: acceptance of genuine responses (CPR, DSR, primary DA, focus in/out) and rejection of non-report escapes like \x1b[2J (clear screen).

No production code changes — coverage only. All 11 tests in the file pass locally (vitest run src/lib/terminal-report-filter.test.ts).

[cloudflare-workers-and-pages]

Deploying mouseterm with    Cloudflare Pages

Latest commit:	c681bec
Status:	✅  Deploy successful!
Preview URL:	https://edb4f4d3.mouseterm.pages.dev
Branch Preview URL:	https://nightly-report-filter-tests.mouseterm.pages.dev

View logs