The directory's dataset IDENTITY: the content hash its chunk files were
written under. Separate from the index because the index is rewritten on a
debounce and a reload can land before any save of it completes, while this
file changes only when the hash does and is written BEFORE the first chunk
under that hash (OPFSStore.doSet waits for it). So every chunk file in the
directory has a known provenance even when the index never landed.
The directory's dataset IDENTITY: the content hash its chunk files were written under. Separate from the index because the index is rewritten on a debounce and a reload can land before any save of it completes, while this file changes only when the hash does and is written BEFORE the first chunk under that hash (
OPFSStore.doSetwaits for it). So every chunk file in the directory has a known provenance even when the index never landed.