Filter out invalid notations

[nichtich]

Notations given in PICA contain errors, so better filter out syntactically invalid notations. Keeping them in the full TSV files (e.g. rvk.tsv) is useful but further processing should remove them:

• [ ] add BARTOC URI to classifications.csv and extract notationPattern or directly add notationPattern to classifications.csv
• [ ] extend classification-subjects.sh to filter out invalid notations

We may add additional rules for cleanup, e.g. remove / in DDC notations, normalize whitespace...

[nichtich]

Note that cleanup can lead to additional duplicated notations so order of processing is relevant.

[nichtich]

Cleanup may also include looking up whether a notation actually exists, e.g. lnot every pattern ^[0-9][0-9].[0-9][0-9]$ is valid BK. As each vocabulary has its own content and rules, best create a service for each vocabulary.