Update processing to match current server behaviour

[Zastai]

Some of the endpoints seem to be returning data that does not match the current API documentation:

• artist info (e.g. user and sitewide artist stats) has a single artist_mbid instead
• artist map has additional information: for each country, an artists field containing a list of artist info
• the result of the listens endpoint includes a new timestamp field (oldest_listen_ts)
• the mbid_mapping field of a recent listen now has several additional fields (artists, caa_id, caa_release_mbid and recording_name)
  • the artists field is a list of artist credits (artist_mbid + artist_credit_name + join_phrase)
  • artists, caa_id and caa_release_mbid also appear in the data for the recording and release statistics

See also LB-1405.

[Zastai]

In addition, MessyBrainz IDs may now be obsolete - they will be useless to API users (MessyBrainz is now an internal component of ListenBrainz instead of its own user-facing project). Only delete-listens still mentions them, but we don't have that endpoint yet anyway.