'use_s_picks=False' does not work as intended in relative_magnitude()

[darren-tpk]

Describe the bug relative_magnitude() will incorrectly output the delta mag of an S-picked station in the dictionary output IF there exists a P and S pick on the same station in event.picks, despite setting use_s_picks to False.

To Reproduce

• Input an event pair with P and S picks on a common station into relative_magnitude(). (e.g. my event pair has picks on 'NC.MINS..HHZ' and 'NC.MINS..HHN')
• relative_magnitude() calls relative_amplitude(), which calls _get_signal_and_noise(), which calls _get_pick_for_station()
• If the input in relative_magnitude() has 'use_s_picks=False', the same argument will be put into _get_signal_and_noise()
• However, _get_pick_for_station() matches the station name to the picks in the input event, and while it excludes the S-picked channel, it will output the channel's P pick within _get_signal_and_noise(), even though the seed_id in the loop is that of the S pick.
• This introduces an error where the signal and noise is computed on the S-picked channel trace using the P-picked channel pick times.

Expected behavior

• The S pick should be completely ignored.

Desktop (please complete the following information):

• Operating System: macOS 11.5.2
• Python version: 3.8.12
• EQcorrscan version: 0.4.3

Additional context I am trying to get relative_magnitude to work - when computing the weighted delta_mag using weight_by_correlations, I noticed that the delta_mag of a S-picked channel is still included despite setting use_s_picks to False.

[darren-tpk]

At the moment I can exclude S picks in my event pair before feeding them into relative_magnitude as a short term fix, but it still holds that the 'use_s_picks' argument does not work as intended.

[calum-chamberlain]

Thanks for this note @darren-tpk - this sounds like this shouldn't be the expected behavior. It sounds like there needs to be a change in _get_pick_for_station(). I'm not going to have time in the next couple of months to work on this, but it would be great to see a pull request that fixes this.

Thanks again.