Test failure in test_aa_high_precision()

Description

This is in regards to issue #11527 and PR #11564. At first, I didn’t worry too much about the failure as the difference appeared to be small enough to be ascribed to the change in the handling of polar motion introduced in pyerfa 1.7.2. I calculate that differences due to this change should be less than 1 uas and the difference shown in issue #11547 is 8.63433769e-11 degrees or 0.31 uas, but then along came PR #11564 which is evidently showing the following differences:

Azimuth 19.62 uas
Elevation 0.85  uas
Distance 38.87 mm

These seemed excessively large to me, so I downloaded the latest version of SOFA and reran my test program. The results for the time used in the test are:

       Time            Azimuth         Elevation       Distance (m)    DUT1      Xp       Yp
2017-04-06 00:00:00 15.032673509956 50.303110133923 376252883.247239 0.4634343 0.009587 0.384603

Here I am including the EOP values to make sure that we are still on the same page in that regard. The differences between this and the previous version of SOFA that I used are:

Azimuth 0.252 uas
Elevation -0.018 uas
Distance  0.000 mm

Considering the apparent discrepancy, I think it would be worthwhile to revisit the entire 24 hour test. If someone could please run the following script and post the output here, I can take a look at this.

astropy_topo_test.py.txt

[saimn]

Re-pkg_resources: we replaced it with importlib.metadata (#11091)

[brandon-rhodes]

No problem, having the separate issue helps track the change. I’ve resolved it; __version__ should appear with the next release.