This fixes a crash in ExtrinsicMetadataIndexer when an origin object was
removed from the archive but still referenced in Kafka, and applies masking
to any object that may be read from Kafka

Since the release of pytest 8.1, some pytest options are no longer
needed and editable install can be used when running tests using tox.

Related to swh/meta#5075.

Most other journal clients use this configuration key

Celery tasks for indexer were removed so we also need to stop referencing
them in the swh.workers entrypoint.

And replace comment type annotations by explicit ones.

Docstring is mandatory for a celery task function as its value
is inserted in the required description column of the task_type
table in the scheduler database.

Celery task function name is also used as task name (with underscores
replaced by dashes) so ensure task function names match the task types
registered in production scheduler database.

So that we can get rid of indexer task types being created by
swh-schedulers' sql init scripts.

SingleFileIntrinsicMapping.detect_metadata_files was comparing lowercase
versions of filenames with SingleFileIntrinsicMapping.filename variable
value to detect metadata files.

But as SingleFileIntrinsicMapping.filename holds the canonical name of a
metadata file it can contain uppercase characters.

So ensure to compare lowercase versions of both filenames to avoid metadata
files being undetected.

This fixes indexing of Python intrinsic metadata.

It now needs types-click which is indeed a dependency of
swh.core[testing].

Refs. #4733

Related to swh/meta#4960