Manage srt detection with content starting with a BOM

[lunika]

On the Marsha some conversion of srt files were failing. After investigate a bit, the file content was starting with a Byte Order Mark.
You can look at this PR to check what is the problem: openfun/marsha#2604

We manage this case in our code. Are you interested with a PR checking that in pycaption ? If yes I can open it.

Thanks