Detect separator when there are header rows

I work with a lot of ALE files (Avid Log Exchange). This format has a few header rows of plain text followed by tab-separated data. Table Tool always defaults to a comma separator, but I would love to automatically detect the tab. If I delete the header rows, then it correctly detects the tab separator. Sample file attached.

This heuristic might be difficult to implement, so I propose an easier workaround: A preference window where the user can set the default separator when guessing fails. I would guess most users probably work with the same kind of files often.

[All Takes.ale.txt](https://github.com/jakob/TableTool/files/2168303/All.Takes.ale.txt)

Start by reproducing the separator guess with the attached All Takes.ale.txt file, then locate the code path that chooses the default separator. Done means tab-separated data is detected despite plain-text header rows, or users can set a default separator when guessing fails.

Written by the indexing model from the issue text.