Allow for non-mutated user specified doc-ids

[mskarlin]

This was a deep bug. On round-trip serialization of a DocDetails object, if a user doesn't have a doi field to create a deterministic doc_id, then when serializing/deserializing (during validation), the doc_id was mutating. This prevented users from ever setting their own doc_id value.

This round trip serialization happens during the gather_evidence step, when the existing doc_id for each DocDetails object in Docs would then diverge from the doc_id for the associated Context objects generated. So for each call to gather_evidence, if a DocDetails did not have a DOI, a random doc_id was created for each Text.doc object within each Context. This PR fixes this behavior.

As an aside, my tests were failing because the RetractionDataPostProcessor test file (tests/stub_data/test_retractions.csv) was more than 30 days since the creation date. So this triggered a cache break which tried to rebuild the file. This then triggered a network call which broke the VCR cassette cache. I turned this behavior off by default, as I don't think this is needed under normal circumstances.

[Copilot]

Pull Request Overview

This PR fixes a bug where user‐specified doc_ids were being overwritten during round-trip serialization of DocDetails objects. Key changes include:

• Updating the DocDetails validation in paperqa/types.py to preserve user‐defined doc_ids.
• Enhancing test coverage with a new test_docdetails_doc_id_roundtrip to ensure serialization does not mutate doc_ids.
• Modifying clients in retractions and journal_quality to explicitly pass through the doc_id and dockey fields.

Reviewed Changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File	Description
tests/test_paperqa.py	Added tests to verify consistent doc_id behavior upon serialization.
paperqa/types.py	Updated the doc_id population logic to honor user-specified values.
paperqa/clients/retractions.py	Adjusted _process to preserve doc_id/dockey when merging DocDetails.
paperqa/clients/journal_quality.py	Updated merging of DocDetails to preserve doc_id/dockey.

Comments suppressed due to low confidence (1)

tests/test_paperqa.py:1576

• The assertion message contradicts the condition being tested. Update the error message to indicate that the doc_id should match the specified test value.

    ), "DocDetails with doc_id should not match test_specified_doc_id"

[jamesbraza]

Perhaps let's improve these comment a bit, imo in their current form they're just mirroring the code

A good comment should mention why -- perhaps explain why was the word "preservation" used

 DocDetails(
    # Propagate doc_id and dockey to preserve them
    # so DocDetails construction doesn't generate new ones
    doc_id=doc_details.doc_id,
    dockey=doc_details.dockey,

Or maybe remove the comment entirely

[jamesbraza]

if not data.get("doc_id") may be simpler, sometimes the double negatives are harder to comprehend