[MNT] Rewrite API Parser and Data Loading

[theOehrly]

The API parser and data loading/processing code have reached a point where they should be rewritten from scratch, taking into account what has been learned over the last years. Too many things have been added in to fix problems that have come up. The code flow is convoluted and fragile. The following is a non-exhaustive list of things that need to be done.

Must be done

• rewrite API parser
• remove 150s lap time limit ([BUG] Restart lap times after the red flag #473, [BUG] Duplicate lap times in source data for GRO incorrectly handled in 2020 Round 4 #612, [BUG] Timing data contains laps with incorrect duplicate lap times #404)
• drop usage of Pandas in API parser, switch to all numpy Datetime and Timedelta (Audit all assignments to pd.NaT #728)
• move post-processing code out of Session class

Optional

• allow users to configure what post-processing is done
• drop the F1 Livetiming first approach, make data sources "equal" in importance thereby allowing to load all lap times from Jolpica, for example. Make data source priorities configurable.
• update cache implementation
• enable dynamic data loading (?)