fix: handle zero-length dimensions in HDFParser for zarr-python 3.2.0

[TomNicholas]

Summary

• Zarr v3 allows shape=(0,) but forbids zero-length chunk dimensions, now enforced by zarr-python 3.2.0+ (see zarr-developers/zarr-python#3711).
• HDFParser previously fell back to dataset.shape for the chunk shape when dataset.chunks was None, producing an invalid chunk_shape=(0,) for empty HDF5 datasets.
• Fix clamps each chunk dim to >= 1, so empty HDF5 datasets round-trip cleanly on both zarr 3.1.x and 3.2.0+.

Unblocks the test_empty_dataset failure noted in #957.

Test plan

• test_empty_dataset passes on zarr 3.1.6
• test_empty_dataset passes on zarr >=3.2.0 (verified that zarr accepts shape=(0,), chunks=(1,) on both dev and released builds; full upstream-zarr test run blocked until Add compatibility with Zarr-Python 3.2.0 #957 lands and provides the RegularChunkGridMetadata import shim)
• No regressions in virtualizarr/tests/test_parsers/test_hdf/

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.31%. Comparing base (b3899e1) to head (88c3d8c).
⚠️ Report is 3 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #977   +/-   ##
=======================================
  Coverage   89.31%   89.31%           
=======================================
  Files          33       33           
  Lines        2031     2031           
=======================================
  Hits         1814     1814           
  Misses        217      217           

Files with missing lines	Coverage Δ
virtualizarr/parsers/hdf/hdf.py	95.58% <100.00%> (ø)

... and 1 file with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.