fix download mirrors

[mtran-code]

Summary

• Replace legacy CCLE 2019 DepMap portal download links with reachable Broad static mirror URLs.
• Add a shared workflow/scripts/download_resource.py helper that downloads or copies resources atomically, rejects HTML verification/error pages, and validates gzip/zip content where expected.
• Route download rules through the helper so gated DepMap pages fail loudly instead of being saved as raw CSV/GCT inputs.

Root cause

The DepMap portal download endpoints can return an HTML verification page with HTTP 200. Rules using raw curl -L could either fail later with gunzip: unknown compression format or silently save HTML into raw data outputs.

Validation

• uvx ruff format workflow/scripts/download_resource.py
• uv run python -m py_compile workflow/scripts/download_resource.py
• Validated 17 updated https://data.broadinstitute.org/ccle/... mirror URLs with curl -fsSI; all returned reachable non-HTML responses.
• Real helper smoke tests:
  • downloaded Cell_lines_annotations_20181226.txt through the helper.
  • downloaded and decompressed gencode.v19.genes.v7_model.patched_contigs.gtf.gz through the helper.
  • confirmed a DepMap verification-page URL is rejected by the helper.
• uvx --python 3.11 --from snakemake==7.32.4 --with pulp==2.7.0 snakemake --dryrun --cores 1
• uvx --python 3.11 --from snakemake==7.32.4 --with pulp==2.7.0 snakemake --cores 1 -F metadata/Cell_lines_annotations_20181226.txt rawdata/epigenetics/methylation/enhancer_cpg_clusters.tsv
• git diff --check

Notes

The remaining DepMap portal URLs are current/public-release files without confirmed static mirrors in this pass: OmicsSomaticMutations.csv, OmicsCNGeneWGS.csv, OmicsCNSegmentsWGS.csv, and CCLE_miRNA_MIMAT.csv. These now fail with an explicit validation error if the portal returns a verification page.