Enhancing `TimeSeriesSource`s with method to extract future activation ticks

[staudtMarius]

Resolves #543 Resolves #572

[sebastian-peter]

Note to myself: Does it make sense to only retrieve the next tick after the current instead?

Edit: Considering the principle of discrete-event simulations, we only need the very next tick. But since a tick as a Long only takes 8 bytes of memory, we only have ~ 70 kbytes for a whole year of hourly-spaced ticks (for one time series). This does not seem like a problem (currently).