Bio2zarr

- Fix bug in handling all-missing genotypes (328)

- Fix missing dependency issue for packaging

- Support out-of-order field definitions in the VCF header (322, ACEnglish)

- Reduce memory requirement for encoding genotypes with large sample sizes

- Transpose default chunk sizes to 1000 variants and 10,000 samples (issue:300)

- Add chunksize options to mkschema (issue:294)

- Add experimental support for local alleles.

- Add experimental support for ``region_index``

Breaking changes

- ICF metadata format version bumped to ensure long-term compatility between numpy 1.26.x
and numpy >= 2. Existing ICFs will need to be recreated.

Maintenance release:

- Pin numpy to < 2
- Pin Zarr to < 3

- Initial production-ready version.
- Add -Q/--no-progress flag to CLI
- Change num-partitions argument in dexplode-init and dencode-init
to a named option.

- Change output format of dexplode-init and dencode-init
- Bugfix for mac progress, and change of multiprocessing startup strategy.