feat(beholder): capture console messages and page errors

[YusukeHirao]

Summary

• Capture console.log/warn/error/etc. and uncaught exceptions / unhandled promise rejections during page scraping, for internal pages only (isExternal: false), via page.on('console') / page.on('pageerror').
• Return them as ScrapeResult.consoleLogs (ConsoleLogEntry[]), mirroring the existing resources/pageUrl pattern rather than embedding into PageData — matches how sub-resources are already collected as a page-scoped array, not a per-page-data field.
• #fetchData's navigation/extraction body is now wrapped in try { ... } finally { await Promise.all(pendingConsoleWork); }, so async jsonValue() argument resolution is guaranteed to settle before the function returns or throws (previously-considered per-return-statement awaits would have missed the two thrown-error exit paths and let a @retryable retry contaminate the array with a stale attempt's late resolutions).
• ConsoleLogEntry is exported from the package's public entry point alongside ResourceEntry/NetworkLog.

Scope: @d-zero/beholder only. This is a producer-side addition (ScrapeResult gains a new required field); no consumer in this repo constructs ScrapeResult literals, so nothing else in tools needed updating. A downstream consumer (nitpicker, separate repo) does construct ScrapeResult-shaped literals in a few places and will need a small follow-up there — tracked separately, out of scope for this PR.

Test plan

• yarn build (via NX_WORKSPACE_ROOT_PATH) — no type errors
• yarn lint — 0 errors (pre-existing unrelated warnings only)
• yarn test — 113 files / 1612 tests pass, including new to-console-log-entry.spec.ts / to-page-error-entry.spec.ts covering: per-type propagation, mixed success/failure args() resolution, dispose() invocation, location presence/absence, and pageerror for both Error and non-Error throw values
• Note: scraper.ts's own listener wiring (onConsole/onPageError, the try/finally flush) has no direct unit test — Vitest's esbuild-based transform cannot parse this file due to the pre-existing @retryable decorator on a private method (#fetchData); confirmed by reproducing SyntaxError: Private field '#fetchData' must be declared in an enclosing class when attempting to add a spec file. This is a pre-existing toolchain limitation (no spec file has ever imported scraper.ts), not introduced by this change.