Drift scan observation time

[jpober]

I've become aware of a bug in the calculation of "coherent" observation time for a drift scan observation. It should be the time it takes for the field of view to cross the FWHM of the primary beam, which is frequency dependent, but as it stands, this is only calculated at 150 MHz. While the fix is straightforward, it will require running mk_array_file.py and generating an independent uv coverage for every redshift, which is significantly clunkier than things are now. So I'm having a hard think about more graceful solutions.

In previous results, this bug would (falsely) cause a small increase in sensitivities at high frequencies (overestimating the amount of coherent vs. incoherent integration), and likewise, cause a small decrease in sensitivities at low frequencies. It would not affect tracked calculations, as there the integration time is determined independently of the beam.