Create proper test dataset for background subtraction

[RobJY]

We're currently just using the same tonsil images that we use for other tests. We need to use images with proper background / antibody control cycles.

[kbestak]

To address this, I can provide a small crop of our COMET data (already publicly available), or if we can use larger data exemplar-001 has the autofluorescence channels in cycle 1.