(feat) added auto handling of traditional pin and testing

[jspaezp]

This PR fixes backwards compatibility with 'traditional' pin files (file extension == ".pin", tab delimited, tabs as protein separators, last column == "protein(s)?" caps insensitive).

Basically, when detecting that its a 'traditional' pin, in memory it 'fixes' the separators in the protein column and moves forward with the fixed dataframe. This in practice means that pin files with ragged edges don't benefit from the streaming dataset benefits but are handled transparently in the background (in other words: if your data is small, no need to convert it. If it is large it will break IF it has tabs as separators in the protein column, which I believe its a good compromise)

This PR also incorporates the changes in https://github.com/wfondrie/mokapot/pull/125 (sorry for the large auto-formatted changes).