[BUG] evals['qa_evaluator'].label is None is not the opposite of evals['qa_evaluator'].label is not None

[firetix]

Describe the bug
I'm unable to pull the spans that never been evaluated from Phoenix. In the UI or python if I filter by evals['qa_evaluator'].label is not None it will return the one that are correctly evaluated, but if I do evals['qa_evaluator'].label is None it doesn't return the spans that are not evaluated. How can I retrieve the spans that have never been evaluated, I currently can only retrieve the one that have been evaluated, I could do a script filtering but that won't be optimal.

Thanks for looking at this

To Reproduce
Steps to reproduce the behavior:

1. Add a few examples of QAConversation chain
2. Log evaluation for some of them
3. Try to filter by evals['qa_evaluator'].label is None
px.Client().get_spans_dataframe("evals['qa_evaluator'].label is None ")

Expected behavior
Return the spans that never been evaluated.

Screenshots

[mikeldking]

Thanks for the issue @firetix ! Totally understand this usecase and we will address this. Appreciate you bringing it to our attention.