Allow populate with secondary keys

[CBroz1]

Currently, we use the passed attrs of populate as a restriction with the dj func _jobs_to_do

https://github.com/LorenFrankLab/spyglass/blob/master/src/spyglass/utils/dj_mixin.py#L597-L600

This relies on it being a valid restriction of key_source, which only includes primary keys. In a future enhancement, we would allow secondary keys of parents in this field. The need for primary keys is perhaps not transparent in documentation

[samuelbray32]

Just noting, datajoint's populate command also gets the list of jobs in the same way.

If selecting jobs by upstream secondary keys is a feature we want, we will also need to alter the restriction passed to super().populate() here