Increase robustness of calculate_phase_count

[LukasWallrich]

• Document that phases should be called screened and final
• Allow for any capitalisation (tolower(cite_label) == "screened") etc)
• Warn if data does not contain screened and final labels

[TNRiley]

completed

[LukasWallrich]

One addition: can the function work when only final records are identified? Then the screened columns should be dropped - often that seems to be the case, e.g., then comparing search results against a benchmark set, or when only raw and final data are available? Or does this already work?

[TNRiley]

Good call, I can work to strip the screened info if only the final label is applied. In terms of benchmarking, you shouldn't need a label, just to name the source as benchmark.

[TNRiley]

I decided to just make changes to the precision_sensitivity_table, instead of both the count function and the table. The table will now check to see if the screened column is all 0's, if they are, it will filter that column from the table. Checked with data that both includes and does not include screened data and everything looks good.