Dropping irrelevant contexts

[jamesbraza]

This PR clears up some fuzziness in our evidence accounting:

• Removes irrelevant evidence, instead of continually filtering them out
  • Comes with the benefit of much smaller PQASession
• Updates Best evidence(s) message for clarity
  • Including the 'for the current question', to make it clear the best evidence varies per question
  • Using a bulleted list over a numbered list, since order is defined by score (not element index)

[Copilot]

Pull Request Overview

This PR improves evidence management in the PQA (Paper Question Answering) system by filtering out irrelevant evidence contexts early and updating the evidence display format for better clarity.

• Filters contexts with score ≤ 0 at collection time instead of repeatedly filtering during processing
• Updates evidence display to use bullet points instead of numbered lists and clarifies scope with "for the current question"
• Removes test code that artificially manipulated context scores with random values

Reviewed Changes

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

File	Description
src/paperqa/docs.py	Adds filtering to exclude contexts with score ≤ 0 during evidence collection
src/paperqa/agents/tools.py	Updates evidence display format and removes score filtering (now handled upstream)
tests/test_paperqa.py	Removes random import and updates tests to work with new evidence filtering behavior
tests/test_agents.py	Updates test assertions to expect bullet point format instead of numbered lists

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[dosubot]

Related Documentation

Checked 1 published document(s). No updates required.

^{How did I do? Any feedback?}  

[jamesbraza]

Note there was trailing whitespace here previously, now there's not

[dosubot]

Documentation Updates

Checked 1 published document(s). No updates required.

^{How did I do? Any feedback?}