Two bugs during extraction of sub-circuits

[MWolfR]

Affects: brainbuilder.utils.sonata.split_population.split_subcircuit.

1. When extracting a subcircuit with the “do_virtual” option active, no filtering is applied to the virtual populations that are propagated to the extracted circuit. Currently, if one extracts e.g., “Layer4Inhbitory” from a circuit, then virtual populations with “layer” and “synapse_class” properties are filtered as well. That should not be the case.

2. When extracting a subcircuit with the “create_external” option active, the newly created extrinsic edge file should reference the correct presynaptic node population. Currently, it references the name of the node population in the parent circuit, but it should be the new name of the node population, e.g., with “external_” prepended. The entry in question is in the dataset attributes of the sonata edge file at: "edges/$POP_NAME/source_node_id". The property in question is "node_population"