read_fwf based implementation

[e-eight]

Based on the discussion in the previous repository, I created an implementation, for both AME and NUBASE, based on Pandas read_fwf function in the read_fwf branch of my fork. This vectorizes the cleaning up of the data.

I also added the datasets for AME 1993 and 1995. We can also add the older AME/NUBASE datasets.

Feel free to take a look at this implementation, and if you feel it is useful, I can make a PR.