Ecephys pipeline writing extra device and extra electrode groups when setting group_names

[h-mayorquin]

This issue arose during the DiCarlo data upload. Since my last conversion, the names of the electrode groups (in this case, the Intan ports) have been added to the metadata. This addition has caused the NWB file to contain more devices and electrode groups than expected, and the extra ones are incorrect.

Upon investigation, I found this to be a general problem. I will first address the general issue and then implement Intan-specific fixes. Here is a minimal example illustrating the general issue:

from neuroconv.tools.testing.mock_interfaces import MockRecordingInterface


interface = MockRecordingInterface(num_channels=4)

recording_extractor = interface.recording_extractor
recording_extractor.set_channel_groups(groups=[0, 1, 2, 3], channel_ids=[0, 1, 2, 3])   
recording_extractor.set_property(key="group_name", values=["group1", "group2", "group3", "group4"])
nwbfile = interface.create_nwbfile()

assert len(nwbfile.devices) == 1
assert len( len(nwbfile.ec_electrode_groups) == 4

Both fail on main (the results being 2 and 8).\

I will push both fixes soon.