Coarse-graining creating mismatch between frequency bins
When running pygwb on magnetometer data, I was experimenting with greater values for the segment_duration and I encountered an issue for mismatching shapes of arrays, in particular the array of S_alpha and orf when computing the dsc. This mismatch was due to the slightly different frequency arrays used to compute both quantities. Apparently, after some more tests, this happens when segment_duration*frequency_resolution is not an integer. In that case, coarse-graining cannot handle this properly it seems and creates CSDs with slightly different frequency values than we need. Those values are then used to compute ORF as well. I propose we do some kind of manual checking if the frequency values are correct.