Avoid sequence id parsing

[arnkress]

Greetings,

The program correctly detects and parses Uniprot header, as described in https://github.com/soedinglab/mmseqs2/wiki#identifier-parsing

Is there a way to avoid parsing it and keep the entire FASTA sequence identifier (e.g. sp|Q8I6R7|ACN2_ACAGO instead of only Q8I6R7) in the result file when doing a search ?

[milot-mirdita]

#557 is the same issue. We'll think of something.