test: OpenSearch - reorganize test suite

[anakin87]

Related Issues

While working on #1562, I realize that the OpenSearch test suite had become complex, disorganized, and difficult to manage.

Proposed Changes:

• move all document store fixtures to conftest.py, document their purpose, unify sync and async versions, and remove a redundant fixture
• move authorization-related tests from test_document_store.py to test_auth.py
• move filter-related tests from test_document_store.py to test_filters.py
• move async tests from test_document_store.py to test_document_store_async.py (new)

How did you test it?

CI

Notes for the reviewer

This PR is based on the branch of #1562, so please review but do not merge until #1562 has been merged.

Checklist

• I have read the contributors guidelines and the code of conduct
• I have updated the related issue with new insights and changes
• I added unit tests and updated the docstrings
• I've used one of the conventional commit types for my PR title: fix:, feat:, build:, chore:, ci:, docs:, style:, refactor:, perf:, test:.

[julian-risch]

Looks quite good to me already. Maybe it's out of scope for this reorganization of the test suite but I noticed that we repeatedly define the same list of documents to write them to document stores. Let's quickly discuss before merging this PR.

• test_bm25_retrieval_with_filters
• test_bm25_retrieval_with_custom_query
• test_bm25_retrieval_with_custom_query_empty_filters
  use the same list of documents.

Similarly the following tests have the same documents in common:
test_bm25_retrieval_all_terms_must_match_false
test_bm25_retrieval_with_fuzziness
test_bm25_retrieval_all_terms_must_match
test_bm25_retrieval

What do you think about removing duplicate lines of code by defining these lists of documents only once in conftest.py or even a document store fixture that already has all these documents indexed. The latter could be an option because the tests listed above only read from the document store. However, it would at some complexity and speed up is negligible so I'd prefer the the first option of simply defining two lists of documents in conftest.py. Should be as simple as adding the following to that file, right?:

@pytest.fixture()
def test_documents():
    return [
        Document(content="Haskell is a functional programming language"),
        Document(content="Lisp is a functional programming language"),
        Document(content="Exilir is a functional programming language"),
        Document(content="F# is a functional programming language"),
        Document(content="C# is a functional programming language"),
        Document(content="C++ is an object oriented programming language"),
        Document(content="Dart is an object oriented programming language"),
        Document(content="Go is an object oriented programming language"),
        Document(content="Python is a object oriented programming language"),
        Document(content="Ruby is a object oriented programming language"),
        Document(content="PHP is a object oriented programming language"),
    ]

What do you think?

[anakin87]

@julian-risch great idea...
I applied it in bc573d2, and now we are removing a lot of code!

[julian-risch]

LGTM! 👍 1100+ lines of code removed. 💪