Incorrect position errors

[cdeil]

I just saw this in

pos:
  ra: {val: 19h43m55s, err: 1s}
  dec: {val: +21d18m8s, err: 17s}

Note:

In [1]: from astropy.coordinates import Angle

In [2]: Angle('17s')
Out[2]: <Angle 17.0 arcsec>

In [3]: Angle('1s')
Out[3]: <Angle 1.0 arcsec>

So s is parsed as arcsec, i.e. correctly for errors given on dms.
But usually erorrs on hms are in "seconds of hourangle", which is 15 arcsec:

In [5]: Angle('0h0m1s')
Out[5]: <Angle 0.0002777777777777778 hourangle>

In [6]: Angle('0h0m1s').arcsec
Out[6]: 14.999999999999998

This can be reviewed and fixed in a systematic way by computing ellipticity of position errors.
Usually position errors are pretty circular.

[pdeiml]

I would prefer to write it e.g. in this way:

pos:
ra: {val: 19h43m55s, err: 0h0m1s}
dec: {val: +21d18m8s, err: 0d0m17s}

The it is written in the same way as the value and misunderstanding of "s", "arcsec" and so on is avoided.

[cdeil]

@pdeiml - 👍 to your suggestion of using "hms" and "dms" format for position errors to mirror the position values.

(Exception: There are a few papers where positions and errors are given in deg, e.g. glon = 42.024 deg +/- 0.123 deg. In those cases, I would suggest to directly enter this, i.e. follow the general philosophy in gamma-cat to enter data as directly as possible, without manual and error-prone conversions / re-formatting, if possible.)

[cdeil]

This was mostly fixed up and made uniform in #82.
In the end I decided to just use the shorter s and m for arc minutes and arc seconds,
because I find strings like "0d0m3s" hard to read and look too similar to 0h0m3s".
So hms is used, but only for RA errors.

The remaining action item here is to plot all position error ellipses and check for outliers, just in case there's remaining data entry issues in the input files. I'll do that at some point, not high priority.