Brings DOIs and PMIDs into one canonical form so the same paper always maps
to the same cache entry and study folder: surrounding whitespace is removed,
https://doi.org/, doi: and PMID: prefixes are stripped, and DOIs are
lower-cased (DOIs are case-insensitive).
Examples
normalize_identifier(c(" https://doi.org/10.1000/ABC.1 ", "PMID: 12345", "doi:10.1/X"))
#> [1] "10.1000/abc.1" "12345" "10.1/X"