cleaner job manager API for initializing the job database

[soxofaan]

brainstorm PR for #635 to have cleaner job db initialization

[jdries]

Already looks good, we could to a factory method like:
job_db = create_job_db(output_file,df)
And then 'guess' the tye of db by looking at file extension or else make it more explicit.
The difference with:
job_db = CsvJobDatabase(output_file).initialize_from_df(df)
is however minimal, so this version is also just fine for me.

[soxofaan]

yeah I was considering something like that, e.g.:

job_db = CsvJobDatabase(output_file, initialize_from=df)

But then I figured that splitting it in a separate method allows the user to do the following in their scripts:

job_db = CsvJobDatabase(output_file)
if not job_db.exists() or some_other_condition():
    job_db.initialize_from_df(df=get_df())

giving them the freedom to do the initialization in the way that makes most sense to them