Remove averaging of positions in MCs.

[aleberti]

Closes #135

[jsitarek]

the code was also pruning the subarray of the telescopes that are not in use (i.e. have assigned id of 0). For the MAGIC+LST1 and MAGIC-only I think this is fine because those subarrays are already created as they shoud.
But for the simulations with multiple LSTs, or for MAGIC-mono case (i.e. e.g. M2+LST-1 observations), the code I think will introduce changes w.r.t. previous behaviour. Might be useful if @ranieremenezes and @Elisa-Visentin have a look as well.

[aleberti]

@jsitarek This is still done. The relevant part is:

event_source = EventSource(
        input_file,
        allowed_tels=list(
            filter(lambda check_id: check_id > 0, assigned_tel_ids.values())
        ),  # Here we load the events for all telescopes with ID > 0.
        focal_length_choice=focal_length,
    )

by passing allowed_tels, internally the SimTelEventSource will select the telescopes from the subarray, and the lambda function is already taking away those with telescope ID set to. See https://github.com/cta-observatory/ctapipe/blob/main/src/ctapipe/io/simteleventsource.py#L696-L697

[jsitarek]

ok, very good, thanks for checking