I found some infelicities while investigating #1826. This test was hard-coding
a specific path for bash-completion, which was the wrong one for my system, so
it was being skipped every time. This searches the list of paths used by the
actual command.
* Rename the importer argument and related variables to make it more
generic, as the feature should be independent of the backend used and
not restricted to MusicBrainz.
* Update documentation and docstrings accordingly.
* Add changelog entry.
* Store the user-supplied MusicBrainz IDs (via the "--musicbrainzid"
importer argument) on ImporTask.task.musicbrainz_ids during the
lookup_candidates() pipeline stage.
* Update test cases to reflect the changes.
* Fix an issue that caused the candidates for a singleton not to be
returned ordered by distance from autotag.match.tag_item(), when
searching multiple MusicBrainz ids (ie. several "--musicbrainzid"
arguments). The candidates are now explicitely reordered before being
returned and before the recommendation is computed.
* Fix test_importer.mocked_get_recording_by_id so that the artist is
nested properly (and as a result, taken into account into the distance
calculations).
* Replace the entities used on ImportMusicBrainzIdTest mocking the calls to
musicbrainzngs.get_release_by_id and musicbrainzngs.get_recording_by_id instead
of querying MusicBrainz.
* Other cleanup and docstring fixes.
* Add tests for the "--musicbrainzid" argument (one/several ids for matching
an album/singleton; direct test on task.lookup_candidates() for
album/singleton).
* Add "before_choose_candidate" unit tests (PromptChoicesTest), containing
tests for checking the addition of choices to ui.input_options (album and
singletons), conflict resolution, and callback handling (regular and with
return value).
