Prior to this commit, this was passing along real instance.  This did
not bode well with the __init__ call which expects configuration dict.

Possibly there to satisfy tests so to remove.

Indexer storage expects identifier as bytes

Reviewers: #reviewers, ardumont

Reviewed By: #reviewers, ardumont

Subscribers: ardumont, swh-public-ci

Differential Revision: https://forge.softwareheritage.org/D707

Make the OriginMetadataIndexer fetch rev metadata from the storage instead of getting them via the scheduler.

Summary: Depends on D704

Reviewers: #reviewers, ardumont

Reviewed By: #reviewers, ardumont

Subscribers: swh-public-ci

Differential Revision: https://forge.softwareheritage.org/D705

Reviewers: #reviewers, ardumont

Reviewed By: #reviewers, ardumont

Subscribers: swh-public-ci

Differential Revision: https://forge.softwareheritage.org/D704

related to T1330

Summary: Also fixes docstrings.

Reviewers: #reviewers, ardumont

Reviewed By: #reviewers, ardumont

Subscribers: ardumont, swh-public-ci

Differential Revision: https://forge.softwareheritage.org/D702

Summary:
I made several choices while writing that Diff that are open to discussion:

- Using GIN instead of GiST. That seems the most appropriate choice when reading https://www.postgresql.org/docs/9.1/textsearch-indexes.html ; but these part got removed from the doc of pgsql 10: https://www.postgresql.org/docs/10/textsearch-indexes.html
- Using `pg_catalog.simple` as dictionary. Since we're dealing with any language and proper names, it seemed best to use a dictionary with no stop word. Though, arguably, most of the data will be English, and stop words usually don't appear in names.
- It only supports conjunctions of search terms. I could easily add support for arbitrary levels of nestings and disjunctions/negations. That can be done later if we deem it worth it.
- It indexes JSON keys too. It is probably possible to fix this, at the expanse of complicated SQL code, or some postprocessing in Python.

Resolves T1334 and T1335.

Test Plan: tox

Reviewers: #reviewers, ardumont, olasd

Reviewed By: #reviewers, ardumont, olasd

Subscribers: olasd, ardumont, swh-public-ci

Maniphest Tasks: T1335, T1334

Differential Revision: https://forge.softwareheritage.org/D658

Related T1326

Related T1326

And also be consistent with docstring.

Move the decoding steps in test where it belongs

Related T991

So that tests that requires tests fixture without hypothesis does not
break.

Related T1354

- mimetype: no filtering is done.
- fossology_license: Filtering to only compute on textual contents.

Related T1312

The debian stable packaged version hypothesis 3.6.1-1 holds a bug that
prevents the indexer tests from running [1]. That's fixed in earlier
version.

[1] https://github.com/HypothesisWorks/hypothesis/issues/290

Related P334