This should fix ATLASRECTS-7285
The problem was that this algorithm was scheduled twice, and the first case was too early and so data dependencies did not exist. Apparently this instance 'won' in serial, and so muons failed in serial.
I'm not 100% sure whether then other instances of MuonSegmentCnvAlgs in MuonSegmentFindingConfig should also move, but since q445 serial now runs, as do local ctests, I'm inclined not to test this right now.