DOC: include estimated accuracy of built-in solar system ephemeris calculations

According to the SOFA docs page, the accuracy of the default algorithm in astropy calculations is:

**  5) The algorithm is due to J.L. Simon, P. Bretagnon, J. Chapront,
**     M. Chapront-Touze, G. Francou and J. Laskar (Bureau des
**     Longitudes, Paris, France).  From comparisons with JPL
**     ephemeris DE102, they quote the following maximum errors
**     over the interval 1800-2050:
**
**                     L (arcsec)    B (arcsec)      R (km)
**
**        Mercury          4             1             300
**        Venus            5             1             800
**        EMB              6             1            1000
**        Mars            17             1            7700
**        Jupiter         71             5           76000
**        Saturn          81            13          267000
**        Uranus          86             7          712000
**        Neptune         11             1          253000

It would be nice to have these numbers immediately shown in the docs to evaluate whether it is necessary to use more accurate methods. That was my first question on getting to the page, as I need about a few arcsec for Jupiter.

[saimn]

@jgf5013 - Are you using matplotlib 3.1.0 ? This would explain the slowness when building the docs, especially for uncertainty/index. This page took 30min and 9.5Gb of RAM for me with matplotlib 3.1! This will be fixed in their next release.

[saimn]

@pllim :

I think @saimn has a way to smoke out the slowest running doctest ?

Not really, sadly. There is the --durations flag which reports the slowest tests, it works for doctest in rst files as well but reports only the total time for the files:

❯ pytest --remote-data=any --doctest-rst docs/ --durations=10 -vv
...
================================== slowest 10 test durations ==================================
9.36s call     docs/wcs/wcsapi.rst
8.78s call     docs/coordinates/frames.rst
4.92s call     docs/timeseries/index.rst
0.88s call     docs/visualization/normalization.rst
0.66s call     docs/units/decomposing_and_composing.rst
0.58s call     docs/time/index.rst
0.57s call     docs/coordinates/performance.inc.rst
0.51s call     docs/coordinates/index.rst
0.28s call     docs/timeseries/bls.rst
0.22s call     docs/coordinates/velocities.rst
================================= 200 passed in 30.92 seconds =================================