generatemutfasta 1.3: multi-transcript neoantigen peptides + resilience fixes

[johnoooh]

Adds generatemutfasta/1.3 with a --multi_transcript mode that builds neoantigen peptides for alternate transcripts from Genome Nexus' Additional_Transcripts column (-m extended).

Changes

• New optional outputs: alt_mut_fasta, alt_wt_fasta, transcript_map (only when --multi_transcript is set).
• Additional_Transcripts parser (field order Transcript_ID,Hugo_Symbol,HGVSp_Short,HGVSc,Variant_Classification; MAF Variant_Classification vocabulary); canonical transcript emitted from the main MAF columns.
• Primary-path resilience (from a large production run): per-variant try/except skip+count; versionless offline HGVSc requery; guard on missing protein consequence.

Tests

• pytest 41/41 on the parser + resilience logic.
• nf-test 3/3 with --profile docker (real-data + two stub tests incl. --multi_transcript).

Notes

• Container reused: neoantigen-utils-base:1.4.0 (no new image).
• nf-core lint shows 2 failures + 2 warnings that are byte-identical on the existing 1.2 module (meta-name/process-name convention; test-tag guidelines; ghcr unreachable) — carried over at 1.2 parity, not introduced here.
• Real-data --multi_transcript nf-test data is deferred until Genome Nexus -m extended output is released (stub + unit coverage in the meantime).

[johnoooh]

Closing in favor of extracting the generateMutFasta script + its test suite into its own standalone Python package repository (with proper CI), rather than carrying it as resources/usr/bin inside the module. The nf-core module will be revisited to install the packaged script into the container. Branch feature/generatemutfasta is retained for reference.