Compute cross-channel signal statistics for the _aggregate_ regions represented by different-timepoint mergers

[vreuter]

With #279 implemented, we have trace IDs which consist of different regional barcodes'/timepoints' spots. The first implementation of the signal analysis component (#337) considers only actual ROIs, though, treating these aggregate regions component-wise rather than as a collective. We should preserve that regional-spot-specific information but also compute the cross-channel statistics for the "megaregion" (1 trace ID of multiple ROIs).

@TLSteinacker @ines-prlesi what are your thoughts about how you'd like this output structured?

[TLSteinacker]

We could either generate an identical additional output file were the new merged traceId intensities are calculated as the mean intensity of all their merged traceIds, or could otherwise add the new traceId rows in the original output file (this might however be unfavourable if we want to get an overall idea of the measured intensities as it will contain a mixed population).

[TLSteinacker]

After a brief troubleshooting session with @ines-prlesi, we also thought that it might be worth to not only test calculating the mean intensity of the merged traceIds, but to also generate an output file where only the max intensity (of the merged traceIds) is kept. This could be especially important if the IF signal is rather dim and very localised and potentially only present in a subset of the to-be-merged traceIds.