Update metadata embedding code for cell IDs etc

[tmchartrand]

This is necessary to solve the major problem that patch-seq data on different archives uses different IDs - DANDI uses session IDs (ephys roi result ids), the only ones recorded in NWBs currently, while other archives use cell specimen IDs.

Ben Dichter updated the core NWB schema in response to this concern. IPFX just needs to update its pynwb dependency and take advantage of the new functionality: NeurodataWithoutBorders/nwb-schema#512

Additionally, its worth considering using the ndx-dandi-icephys NWB extension to embed additional metadata in a fixed schema: https://github.com/catalystneuro/ndx-dandi-icephys/blob/master/src/spec/create_extension_spec.py
Possibilities include layer, slice ids, cortical depth/thickness, recording temp, and transcriptomic details. The latter might require breaking the metadata embedding code out as a separate utility however, since that info is not stored in LIMS (I don't think).