Dual submission: Use sparse scores and add support for paper subset

[enrubio]

This PR:

• Updates compute_dual_submission_metadata so that it uses the sparse scores and no longer uses heaps to compute the top 5 scores per row.
  • Removes sparse_value as a param since it's now only used to request to the expertise. I think the default for this function should be to compute between active papers using sparse=5 and if PCs want anything else, they should call the expertise separately.
• Updates request_paper_similarity so that you can pass a list of submissions for entity A or B. Useful if you need to compute scores when the venue is already over.

Ran test jobs and checked that output is as expected.

[melisabok]

Adding several optional parameters is very confusing. Let's create three different functions:

	•	request_paper_similarity_by_venue(...)
	•	request_paper_similarity_by_invitation(...)
	•	request_paper_similarity_from_submissions(...)

They can all call to the same inner functions that understand the expertise API request schema:

def _request_paper_similarity_core(
    name,
    entity_a,
    entity_b,
    model='specter2+scincl',
    sparse_value=400,
    baseurl=None,
)

[enrubio]

I see, so you mean request_paper_similarity_by_venue will just call to request_paper_similarity with the correct params?

[melisabok]

This is what I propose:

def request_paper_similarity_by_venue(
    self,
    name: str,
    venue_id: str,
    alternate_venue_id: str | None = None,
    model: str = 'specter2+scincl',
    sparse_value: int = 400,
    baseurl: str | None = None,
):
    return self._request_paper_similarity_core(
        name,
        source_a={'venue_id': venue_id},
        source_b={'venue_id': alternate_venue_id} if alternate_venue_id else None,
        model=model,
        sparse_value=sparse_value,
        baseurl=baseurl,
    )


def request_paper_similarity_by_invitation(
    self,
    name: str,
    invitation: str,
    alternate_invitation: str | None = None,
    model: str = 'specter2+scincl',
    sparse_value: int = 400,
    baseurl: str | None = None,
):
    return self._request_paper_similarity_core(
        name,
        source_a={'invitation': invitation},
        source_b={'invitation': alternate_invitation} if alternate_invitation else None,
        model=model,
        sparse_value=sparse_value,
        baseurl=baseurl,
    )


def request_paper_similarity_from_submissions(
    self,
    name: str,
    submissions,
    alternate_submissions=None,
    model: str = 'specter2+scincl',
    sparse_value: int = 400,
    baseurl: str | None = None,
):
    return self._request_paper_similarity_core(
        name,
        source_a={'submissions': submissions},
        source_b={'submissions': alternate_submissions} if alternate_submissions else None,
        model=model,
        sparse_value=sparse_value,
        baseurl=baseurl,
    )