feat: kind: repo-entity page type for wiki_lint.py

[purisev]

Implements #3.

Adds an opt-in kind: repo-entity page marker for inventory-anchor pages
(e.g. a stub page per git repo listing origin/web_url/description) that are
by design cited only from index.md and carry no other content.

• Orphan check exempts pages with kind: repo-entity (index-only citation
  is the accepted shape for this kind).
• Frontmatter check uses a reduced required set for these pages
  (kind,type,name,origin,web_url,description) instead of the standard
  type,title,tags,created,updated.
• Both behaviors are configurable/disableable via --repo-entity-required-fm
  and --no-repo-entity-exemption, and are backward compatible — a wiki
  with no kind: repo-entity pages sees no change in output.

Smoke-tested: a kind: repo-entity page with no inbound wikilinks and only
kind/type/name/origin/web_url/description in frontmatter produces zero
findings by default, and reappears under --no-repo-entity-exemption /
a stricter --repo-entity-required-fm.

[praneybehl]

Thanks for putting this together. I’m closing this PR rather than merging it into v2.

V2 relies on title, tags, and updated for retrieval, filtering, and freshness, while this change intentionally removes those fields from the required page shape. It also does not implement the equivalent Paperclip lint behavior. In the v2 graph, type: entity, kind: repo-entity has no matching ontology mapping and therefore falls through to the default concept node type, which would misclassify repository pages.

The underlying use case is still not concrete enough to justify a special schema exception. If repository inventory becomes a supported workflow, I would prefer to model repositories as standard entity pages—likely kind: repository—while preserving normal frontmatter and adding an ontology mapping, documentation, cross-runtime parity, and tests.

Thank you for exploring the idea and for following the issue-first contribution process.