[Search SDK] Use Sequence type hint instead of List where appropriate

[pamelafox]

• Package Name: azure-documents-search
• Package Version: 11.4.0b11
• Operating System: Mac OS X
• Python Version: 3.11

Describe the bug

The search() function currently has this signature:

vector_queries: Optional[List[VectorQuery]] = None,

That means if I then have code like:

vectors = []
if has_vector:
    embedding_args = {"deployment_id": self.embedding_deployment} if self.openai_host == "azure" else {}
    embedding = await openai.Embedding.acreate(**embedding_args, model=self.embedding_model, input=query_text)
    query_vector = embedding["data"][0]["embedding"]
    vectors.append(RawVectorQuery(vector=query_vector, k=50, fields="embedding"))

And run mypy, I get a typing error:

backend/approaches/retrievethenread.py:102: error: Argument "vector_queries" to "search" of "SearchClient" has incompatible type "list[RawVectorQuery]"; expected "list[VectorQuery] | None"  [arg-type]
backend/approaches/retrievethenread.py:102: note: "List" is invariant -- see https://mypy.readthedocs.io/en/stable/common_issues.html#variance
backend/approaches/retrievethenread.py:102: note: Consider using "Sequence" instead, which is covariant

I think you could use Sequence in this case, as it's not clear to me that you need it to be a List.

See this related issue:
#29370

Additional context

I am going to workaround like this for now:

vectors: list[VectorQuery] = []