feat(api): rename PDF parser option pageMarkdown -> pages (#4352)

* feat(api): rename PDF parser option pageMarkdown -> pages

Names the option after the response field it populates, matching the
blocks convention ({type: "pdf", pages: true} -> document.pages).
pageMarkdown stays accepted as a silent deprecated alias — it was never
documented, typed in any SDK, or in the OpenAPI spec, so only raw-JSON
early adopters could depend on it; pages wins when both are set.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* refactor(api): fold pageMarkdown alias into pages at the schema boundary (local cubic)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(api): read legacy pageMarkdown on serialized options; refresh schema tests (local cubic)

Serialized scrape options bypass schema re-parsing (queued jobs, crawl
option snapshots written pre-rename), so the getter defensively reads
the legacy key when the canonical pages is unset. Snips validation
tests updated to the normalized shape with alias-fold coverage.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* test(api): exercise the canonical pages option in snips happy paths (local cubic)

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Fable 5 <noreply@anthropic.com>