fix: forgot to implement Acquisition.instrument_id merge

[dbirman]

This PR adds the same instrument_id merge logic that we added to Instrument to Acquisition where we forgot to implement it. See #1705

[saskiad]

I'm a bit confused why we need this. For instrument I get it - we're combining instruments, they have independent instrument metadata. We merge them to make the new metadata.
I don't understand the need to merge instrument ids in the acquisition - that should have already been done.

[dbirman]

I'm a bit confused why we need this. For instrument I get it - we're combining instruments, they have independent instrument metadata. We merge them to make the new metadata. I don't understand the need to merge instrument ids in the acquisition - that should have already been done.

When you split an instrument into two or more parts you end up generating multiple pairs of acquisition/instrument metadata, each of which has its own instrument_id.

[saskiad]

but when it's being used, the instruments are together and the scientists have create a merged instrument file. So shouldn't that be the instrument id they enter into their acqusition?

[dbirman]

but when it's being used, the instruments are together and the scientists have create a merged instrument file. So shouldn't that be the instrument id they enter into their acqusition?

If a scientist pre-merged their instruments then yes, but if they use the GatherMetadataJob to merge their instruments then no. See https://docs.allenneuraldynamics.org/en/latest/acquire_upload/upload.html#merge-rules. We support this already for vr foraging + fib / ecephys