Make waveform units configurable

[stanbot8]

Summary

Units(waveform_unit="microvolts") writes volts for all waveform datasets. This change writes the requested unit and keeps volts as the default. From NeurodataWithoutBorders/pynwb#2162.

Design

Use default_value for waveform_mean, waveform_sd, and waveforms. This keeps the existing default and permits the API value.

The implementation changes core/nwb.misc.yaml and docs/format/source/format_release_notes.rst. The design follows Use default_value because this change is reasonable and non-breaking.

Tests

The HDMF validator accepts the complete core namespace. PyNWB writes and reads microvolts for all 3 waveform datasets. MatNWB reads all 3 HDF5 unit attributes.

The PyNWB relevant suite passes with 60 tests and 662 subtests. The complete MatNWB R2025b suite passes. An installed PyNWB wheel passes the custom unit case. The release note is included. The hdmf-common-schema pointer stays at 1.9.0 because 1.10.0 has an unrelated breaking MeaningsTable change.

Downstream validation includes NeurodataWithoutBorders/matnwb and NeurodataWithoutBorders/pynwb.

Commands

uvx --from git+https://github.com/hdmf-dev/hdmf.git validate_hdmf_spec core -m nwb.schema.json
PYNWB_NO_CACHE_DIR=1 python -m pytest tests/integration/hdf5/test_misc.py tests/unit/test_misc.py -q

[h-mayorquin]

This LGTM and is the approach discussed already here: NeurodataWithoutBorders/pynwb#2162