chore(deps): update dependency @nitpicker/crawler to ^0.21.0

[renovate]

This PR contains the following updates:

Package	Change	Age	Confidence
@nitpicker/crawler (source)	^0.18.0 → ^0.21.0

---

Release Notes

d-zero-dev/nitpicker (@​nitpicker/crawler)

v0.21.0

Compare Source

Note: Version bump only for package @​nitpicker/crawler

v0.20.0

Compare Source

Bug Fixes

• crawler: stop cross-host redirect leak; honor fetchExternal in list (3b3cc12)

v0.19.0

Compare Source

Bug Fixes

• crawler: compute page body hash once per page instead of twice (f2fbda5)

• crawler: route self-healing and session notices away from bare console.* (deb5684)

• feat(crawler)!: implement Symbol.asyncDispose for archive lifecycle (de60624)

• feat(crawler)!: replace Wappalyzer-only page_tags with technology detection (4b81a47)

Features

• crawler: assert puppeteer install is shared with beholder (5cbf0ed)
• crawler: eliminate silent stretches across archive I/O (issue #​294) (eabdd07)
• crawler: export setup-phase label sequences for TaskList consumers (793b952)
• crawler: log page_tags row count before migration (#​294) (5f535a7)

BREAKING CHANGES

• requires Node >=24.11 for native Explicit Resource
  Management (using/await using) support.

ArchiveAccessor (and, via polymorphism, Archive) and CrawlerOrchestrator
now implement Symbol.asyncDispose, delegating to their existing
close()/garbageCollect() logic. Callers can use await using instead
of a manual try/finally around teardown.

• page_tags is removed outright in favor of two new
  tables: technology_signals (raw per-signal evidence, one row per
  signal) and page_technologies (noisy-OR confidence-combined roll-up,
  one row per page x technology). Existing archives are converted via a
  read-time migration (migrate-page-tags-to-page-technologies.ts for
  0.13+ archives, a special case in retarget-legacy-fk-tables.ts for the
  0.10->0.13 upgrade path) that drops page_tags after conversion.
  Page.getTags() and the TagsSummary/TagInventoryEntry types are
  removed along with it; use Page.getTechnologySignals() /
  Page.getPageTechnologies() instead.

Also adds a post-crawl JS license-comment scan (scanJsResourcesForTechnologySignals)
that re-fetches the leading bytes of not-yet-scanned internal JS
resources to detect technologies whose signature lives in the JS body
rather than the HTML — a distinct "post-crawl network enrichment"
category from the existing post-hoc-marking and backfill patterns,
since it is the only step that performs network I/O after the crawl
itself finishes. Each resource is scanned at most once (recorded in
technology_js_scan_cache), and a match folds into every page that
references the resource by recombining that page's full signal set
under the same Scoped-Replace invariant insertTechnologies uses.

Also detects Web Components (custom elements, tag names containing a
hyphen and not one of the 8 reserved SVG/MathML names) within the main
content region, at the same grain as the existing button/table/etc.
categories (page_main_content_custom_elements, main_content_custom_element_count).

0.18.1 (2026-08-12)

Bug Fixes

• crawler: close two more demotion sites the /code-review pass on bf75031 found (69c3597)
• crawler: guard is_external against demotion on an already-scraped internal page (bf75031)
• query: exclude out-of-scope pages from the directory tree (56a2c6b)

v0.18.1

Compare Source

Bug Fixes

• crawler: close two more demotion sites the /code-review pass on bf75031 found (69c3597)
• crawler: guard is_external against demotion on an already-scraped internal page (bf75031)
• query: exclude out-of-scope pages from the directory tree (56a2c6b)

---

Configuration

📅 Schedule: (UTC)

• Branch creation
  • At any time (no schedule defined)
• Automerge
  • At any time (no schedule defined)

🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.

♻ Rebasing: Whenever PR is behind base branch, or you tick the rebase/retry checkbox.

🔕 Ignore: Close this PR and you won't be reminded about this update again.

---

• If you want to rebase/retry this PR, check this box

---

This PR was generated by Mend Renovate. View the repository job log.