Unify SQL time series tables of same type.

[sebastian-peter]

Current situation: Each time series exists in its own table (analogous to separate files for csv time series). This means that many tables with the same column scheme can exist.

This structure can be improved by merging all tables of the same type (e.g. time_series_p, time_series_pq etc.) and adding a column specifying the time series uuid. Together with #515, initialization of SQL sources could be sped up significantly, as the advantages of relational databases (indexing etc.) are put to use here.