Replace gbasf2 output string parsing with our own scripts interacting with gbasf2

[meliache]

Assign this to myself, but @philiptgrace was also interested and asked about the string parsing here

Interacting with gbasf2 has to be via subprocesses, there is no other way around, because gbasf2 runs with another environment and another PYTHONPATH than b2luigi. Often, I was just lazy and called the user scripts provided by gbasf2 from the gbasf2 BatchProcess via subprocesses and captured and parsed their output when I needed it.

But it should be more stable to just write for each instance where I need it a short python2 script that directly imports and works with the BelleDirac python package which is available from the gbasf2 environment. Then I could run these scripts as subprocesses and obtain information e.g. via json or returncodes.
It is possible to just import the gbasf2 python packages and work with them.

Places where I could do that replacement of string-parsing via my own scripts

• check if dirac proxy is inititialized instead of parsing gb2_proxy_info here -> PR Use Dirac API for Dirac proxy checking and checking for gbasf2 project existance #51
• for getting the list of filenames in the output dataset instead of using gb2_ds_list here
• for checking if an output dataset even exists on the grid instead of this
• checking if the project already exists on the grid here -> PR Use Dirac API for Dirac proxy checking and checking for gbasf2 project existance #51
• for getting the dirac user in get_dirac_user()