* Decode AVIF through shared image codecs
* Include JPEG encoder distribution notices
* Exercise AVIF HDR decoding through public API
* Preserve AVIF source bit depth
* Test AVIF alpha through public entrypoints
* Update AVIF round trip for supported decode
* Document native AVIF decode support
* Clarify import feature gates
* Keep RAW fixtures valid under shared container and strip bounds
* Include sample-to-chunk mappings in CR3 fixtures
* Find Similar test: a photo read from memory isn't a missing read (#3823)
The second search after a stop plans its reads after frames in which the
loupe and filmstrip decoded photos; those are measured from memory
(source_ref takes the cached source), so the drive sees fewer than 12 reads
on a loaded machine. Check instead that no file is read twice, the searched
photo above all; the answer still needs every photo's signature.
* Find Similar test: every photo is measured after the resumed search
* Bound AVIF padded picture storage and preserve integer source flag
* Size the AVIF picture-store budget for the stored canvas, not the clap-cropped size
* Attribute the AVIF lossless decode fixture
* Keep the AVIF clap test helpers free of unwrap for the codecs lint gate
* lightcraft: keep the newer craft-cms pin after merging main
---------
Co-authored-by: Bruno-Alumn <171815746+Bruno-Alumn@users.noreply.github.com>
Co-authored-by: FredApps <153173854+FredApps@users.noreply.github.com>