Add genome data utilities for homology-based metabolic model reconstruction

[ae-tafur]

Main improvements in this PR:

• Features:

  • downloadGenomeData — new function that retrieves GFF3 (.gff) annotation and protein FASTA (.faa) files for a given NCBI assembly accession (both GCF_ and GCA_ prefixes supported) using the NCBI E-utilities API; skips download if files are already present locally
  • getGeneData — new function that parses the GFF3 annotation file for a given assembly and produces a gene-mapping table with columns locus_tag, old_locus_tag, extra_id, GeneID, gene_name, and GenBank_protein; handles both eukaryote (gene→mRNA) and prokaryote (gene→CDS) annotation styles; calls downloadGenomeData automatically if no local files are provided
  • processProteinFastaFile — new function that renames each sequence header in a protein FASTA file using a user-specified column from the gene table (e.g. locus_tag, gene_name) matched via the GenBank_protein accession; streams the file line-by-line for memory efficiency on large proteomes; sequences with no match in the gene table retain their original header
  • All the functions are updated as the last commit in develop3 branch (Reformat function help blocks to NumPy-style docstrings)

• Motivation:
  In homology-based metabolic model reconstruction, the gene identifiers in the model must exactly match the sequence identifiers in the protein FASTA file used for the BLAST search; these three functions provide a self-contained pipeline to download, map, and reformat genome data so that both artefacts are consistent before reconstruction begins

Instructions on merging this PR:

• This PR has develop as target branch, and will be resolved with a squash-merge.
• This PR has main as target branch, and will be resolved as descriped here.

[edkerk]

Rework genome-data utilities: fix eukaryote protein mapping, add tests

Functional changes only to the genome-data functions from this PR; the
original formatting is preserved where behaviour is unchanged:

• getGeneData now takes the protein accession from each CDS protein_id and
  resolves the owning gene through the Parent chain (CDS->mRNA->gene for
  eukaryotes, CDS->gene for prokaryotes). The previous eukaryote path used
  the mRNA transcript accession (NM_/XM_), which never matches the protein
  FASTA headers (NP_/XP_), so eukaryote renaming silently failed. Removes
  the fragile mRNA-scanning organism detection and the organism-specific
  Phytozome 'extra_id' column; %XX decoding no longer uses urldecode (which
  also mangles '+'); writing the .tsv is now opt-in.
• downloadGenomeData uses the NCBI Datasets v2 API (a single request that
  returns both the GFF and protein FASTA) instead of the deprecated
  Assembly E-utilities, and uses correct error identifiers.
• renameModelGenes: fix the broken warning (hardcoded count, mismatched
  format args, data used as a format string), drop the O(n^2) fallback
  tracking, and leave genes with no or empty mapping unchanged instead of
  silently substituting locus_tag.
• Add genomeDataTests with offline GFF3/FASTA fixtures (eukaryote and
  prokaryote) plus a connectivity-guarded download test.

Verified: getGeneData protein accessions match the protein FASTA headers
100% for both S. cerevisiae and E. coli; all tests pass.

[github-actions]

Function test results

196 tests   176 ✅  37s ⏱️
 22 suites   20 💤
  1 files      0 ❌

Results for commit 295d964.

♻️ This comment has been updated with latest results.

[edkerk]

A final set of changes, to align this PR with recent changes in the develop3 branch, for instance in relation to the format of function_tests, and how optional variables are parsed.

[edkerk]

Looks good, very useful, we should include it in https://raven-docs.readthedocs.io/en/latest/protocol/.