if input data have state = PPQQ or Coherence, then add pol[0] and pol[1]

[straten]

To date, clfd has assumed that pol[0] is equal to the total intensity and searched for RFI only there.
This fix updates clfd to check for the polarization state of the input data and, if it is PPQQ or Coherence, then pol[0] and pol[1] are added together before searching for RFI in the result.

[v-morello]

Hi Willem,

As you already know, clfd was developed for search surveys where I think archives were Stokes I only, not only state is Stokes but there's only one pol channel. If we're going to make it compatible with other archive types, one thing to address beyond the loading of the data (which is fine) is the subsequent mutation, i.e. the application of the profile mask and (more tricky) the spike subtraction; cf. the other methods of ArchiveHandler.

The application of the profile mask should be done across all pols, not just index 0; for the spike subtraction, well, we might have an actual problem here 😄

The values to subtract would be derived from PP + QQ, but then applied to PP and QQ separately, and I don't see how that would be correct. Maybe a good starting point would be to forbid spike subtraction outside of specific cases. If you have ideas, I'll take them. I'll have a think too.

[v-morello]

And clearly, there's an independent problem with the CI workflow too.

[v-morello]

The CI pipeline is fixed now, if you can rebase on master