[Paullo9]

Requested by Paul · project thread

Part of #6

Summary

A multi-page TIFF whose pages do not make one series the reader takes is now refused with a typed, named error,
TIFFError.unsupportedSeries(file:series:), instead of a generic FormatoReadError.unsupported without the file name.
The kinds: an OME-TIFF series; a pyramid of reduced-resolution pages (a page with the reduced bit, or a page one
sampling step below the first, as tifffile pyramidize_series does in its generic rule); N series of pages that differ
in shape, type or layout; and the ImageJ and tifffile-shaped stacks the reader cannot place. No file that reads today
is refused; only the error on paths that already refused changes. The DocC article names the new case.

Migration

Code that caught FormatoReadError.unsupported for these files catches TIFFError.unsupportedSeries instead. An
exhaustive switch on TIFFError adds the new case (the apps only use if case, so they need no change).

Tests

• Red first: TIFFSeriesRefusalTests (new) does not compile on main, as the case does not exist; green on this branch.
• It builds multi-page TIFFs byte by byte: two page shapes, a reduced pyramid, a half-size page without the reduced
  bit, an OME description; each checks the typed case, the file name and the text. A uniform stack still reads.
• tools/test.sh --skip FormatoKitParityTests -j 1: green, no warnings.
• Reviewed before commit (public API): no blockers; wording widened to "shape, type or layout" and the pyramid rule
  aligned with tifffile after the review.

Remaining in #6

• Vendor calibrations (TVIPS, Zeiss SEM, JEOL SightX pixel size): next PR.
• The ImageJ and shaped-series refusal branches carry the new case but have no planted file of their own yet.

🤖 Generated with Claude Code