v0.10.0

Added

• Every file in the published folders that is not on the site is named, with
  the reason and what to do: a kind of file the site does not publish, a
  shortcut, or a document the sync could not export. The content health page
  lists them first under "Not on the site", with a link to each in Drive, and
  the folders the configuration ignores with how much each holds. The job
  summary lists the same files by folder, name, type and reason. See
  ADR-025.

• PDF files in the published folders are published. Each gets a page like a
  Google Doc: an address from its name without .pdf, a permanent link, a
  PDF mark in the sidebar. The page shows the file in the browser's viewer,
  with links to open and download it, and its text, which search indexes and
  index.md serves. A PDF over 25 MB, the most the site can serve, is
  published as its text with a link to Drive; one over 100 MB is only linked.
  An unchanged PDF is not downloaded again. The content health page lists a
  PDF with no readable text, or too large for the site, as incomplete. See
  ADR-027.

• data/docs-index.json gives each document a format: google-doc or
  pdf.

Changed

• A document named with a letter from another alphabet is held back and
  listed with the letters to retype, instead of stopping the sync. A folder
  named so still stops it. ADR-027 supersedes ADR-020 in this.
• @ctcstack/ctcdocs-sync depends on unpdf to read the text of a PDF.
• A document that cannot be exported for a reason of its own — over Google's
  10 MB export limit, downloading turned off, or content conversion refuses —
  no longer stops the sync. It is held back: a document already published
  keeps that version and its address, a new one stays off the site, and the
  rest of the corpus is published. It is tried again on every sync. Failures
  that are not the document's own, and a targeted --file run, still stop the
  run. See ADR-026.
• data/latest-sync-report.json moves to schema version 2. It lists the files
  not on the site, the reasons and the ignored folders, and counts them. The
  first sync after upgrading rewrites it.
• The inventory report lists the ignored folders with their paths and item
  counts.