Rework duplicate detection

[Kunsi]

PR #5 introduces HAFASEvent.id, which could be used to determine if departures are listed multiple times. This would remove the need to have a lot of the logic we currently have for detecting duplicates.

Ideally, we'd have two modes of duplicate detection:
a) Only show the first event (depending on departure/arrival time)
b) Prioritize stops according to their order in stop_ids, so if a journey travels through multiple stops, the event gets prioritized by the order of stop_ids, so listing a stop earlier will prioritize it.

[Kunsi]

For cancelled services with a replacement service running:

If you have two services, scheduled to arrive within the same minute, one with realtime data, one without, the one without realtime data is a duplicate.