Msstitch

Added
- Support for Sage search engine output

Fixed
- Output MS1 quant correctly if there are missing scans in isobaric quant (15ba5883efbf3043995ff222925b5e8467ee5e20)
- Merge protein crash when some protein was not in DB (2b073aa59cec8c3730e785771fd95099f7e55d8f)

Added
- qvality input can be used for adding FDR to PSM table (8ead34b0cf95fab966634e38fb49cd860b109428)
- `--report_seqmatch` for annotation of a PSM table with peptide matches in external fasta files
- `dedupperco` and `deduppsms` to remove duplicate peptides from percolator data and PSM table

Changed
- Trypsinize and storeseq now separate peptides on stop codons, can be ignored if needed (b6629ef8cb012c1e0dcf9a4ad8fd45a1283eef1f)

Fixed
- When percolator peptide has missing PSMs, set remaining PSMs to FDR=1
- When percolator PSM has no peptide, set PSM to FDR=1 (8ead34b0cf95fab966634e38fb49cd860b109428)
- When multiple different isobaric types are passed to storequant, make sure it doesnt crash by storing correctly
- When multiple different isobaric types are in a PSM table, do not crash when parsing denominator patterns

Changed
- Less memory overhead for `msstitch storequant --fullprotein`
- Reproducible "random" shuffling of target-matching sequences for decoy generation with tryptic reverse

Fixed
- Match isoquant from OpenMS to spectra using scan IDs instead of RT (RT can be identical for multiple spectra in merged TIMS data)

Added
- `isosummarize` cmd can use multiple PSM mappings if that is desired behaviour --split-multi-entries
- Peptide table contains bare sequences and start/stop sites
- PSM table filtering by sequence possible using `seqfilt` cmd
- Optional use of N-term-M loss for trypsinization
- Protein table can also be merged using `--no-group-annotation` (will skip protein groups)

Changed
- Full fasta header used in case of doubt about fasta type (swiss/ensembl)
- TIMS (and other) RT stored in minutes
- `storeseq` includes full protein sequence so no need for fasta parsing in subsequential filtering step
- FDR colpattern not strictly needed in merges

Fixed
- MS1 alignment had a 60x too large retention time window (minutes instead of seconds)