fix: conjugation error in matvis wrapper

[r-pascua]

I don't think this is finished, but I wanted to at least get a start on it since I had kept punting on it. There are a few issues I'd still like to address, but I'm a bit too tired today to write good code, so I'll just note the things here:

• I think the default blt_order may have changed from ('time', 'ant1') to something like ('time', 'baseline'). This is functionally the same, but the different name means the big if clause (~line 420 in matvis.py) will be skipped when it shouldn't.
• I think we can also populate the visibility matrix quickly with ('time', 'ant2') ordering by taking the upper triangular indices of the visibility matrix.
• I would like to write some tests that appeal to the simulated visibilities to ensure the conjugation convention is correct. Here are two relatively simple ideas I have:
  • Single frequency, roughly an hour or so obstime with a small array, with a single source transiting zenith. Check that the fringe-rate transform (using a FFT and not an iFFT) peaks at positive fringe-rates.
  • Single time, roughly ten-ish MHz with a small array, with a single source somewhere overhead. We should be able to analytically calculate the delay spectrum peak for a given antenna conjugation.

[steven-murray]

Hey @r-pascua just checking if you have any thoughts/updates to this. It would be nice to have this error fixed before running of IDR3 sims (potentially with matvis due to the mutual coupling)