v0.6.0
HDF5 scale-offset filter (filter id 6) support in both directions, matching the reference library's integer-lossless (ScaleOffset::Integer) and float D-scale (ScaleOffset::FloatDScale) modes (#13, #16).
Datasets compressed with scale-offset by other tools (the reference C library, h5py, MATLAB) now decode instead of failing with UnsupportedFilter(6). Enable on a new dataset with .with_scale_offset(mode); it may be combined with deflate. Verified byte-for-byte against the reference C HDF5 library in both directions.
See the changelog for details.