suggestion: input data files should be read in with tables and units

[gmbrandt]

I have run into this issue twice already and it is hard to track down. I fed in relative astrometry files where the separations were in mas instead of arc seconds. Also, the requirement that the data files have the same columns in the same order every time is error prone. A new user running into these issues would get a soft failure (their orbit fit returns some crazy unconverged parameters) and might get frustrated

Some potential fixes:

1. Throw warnings to the user if the provided separation (in as) is > 2 or 3. Same if all the PA values are < 1 (if for some reason they provided radians).

2. astropy tables support units. We could change file i/o to use astropy tables with units, so that one can use whatever units they want on separation, rv etc. and then internally convert those values to the units used in the code with astropy.units and astropy.constants. Using astropy tables would also make it so that the input data tables do not need to have a consistent column order. However, it would require the column headers to have the same names (i.e. separation is always separation) or something.