Consider alternative strategies for journey-matcher

[haphut]

The APC messages are sent into one topic and the VP messages into one topic per GTFS Realtime data source, so currently per municipality.

There are several strategies of combining APC messages with VP messages, e.g.:

1. Process messages on a first-come, first-served basis. Timestamps are not used. The order of the messages and thus the matching of the messages depends on when the message delivery from the Pulsar cluster to journey-matcher happens.

   If there's a surprising delay in APC messages so that they arrive after the match-triggering, stop-changing VP message and after its APC wait time, they are matched onto a later stop.

   If the journey-matcher is killed and many APC and VP messages become ready to be sent to the journey-matcher while it is restarted, the order of the AP and VP messages is not guaranteed. There might be mismatches.

2. Matching by server-set event timestamp.

   How to handle messages arriving late due to network problems?

3. Matching by on-vehicle APC device event timestamp.

   How to handle messages arriving late due to network problems?