Pipeline:
Enhancement to the existing ww-sra-cellranger pipeline.

Link to references:

• NCBI SRA Run Selector: https://www.ncbi.nlm.nih.gov/Traces/study/
• Existing sample-list input precedent: ww-rnaseq samples_tsv

Context:
@hvenkat94 asked that ww-sra-cellranger accept a project ID (e.g. SRP/PRJNA) and auto-expand it into the per-sample SRA run accessions.

Instead of auto-expanding, this proposes letting the user supply a plain-text accession list (one run accession per line) as an alternative to the inline sra_id_list array. This is exactly what the "Accession List" button in SRA Run Selector produces, so it drops in with no reformatting.

Proposed input (mirroring the ww-rnaseq precedent):

File? sra_id_file
Array[String]? sra_id_list

Array[String] sra_ids = if defined(sra_id_file)
  then read_lines(select_first([sra_id_file]))
  else select_first([sra_id_list])

Then scatter over sra_ids instead of sra_id_list.

Why this approach (and not project-ID auto-expansion)?

1. A CLI path exists but adds a dependency. esearch -db sra -query <SRP> | efetch -format runinfo (NCBI EDirect) works, but EDirect isn't in the getwilds/sra-tools image and runinfo often returns partial/empty results with no error code, silently dropping samples.
2. The metadata doesn't always encode "single cell." Multi-assay studies mix scRNA, multiome RNA+ATAC, bulk RNA, and WXS. In SRP306983 (~70 runs), library_strategy is RNA-Seq for both single-cell and bulk; the only single-cell signal is a submitter-specific library_name convention (_scRNA, _multiome_RNA vs PT#/TS#). No heuristic picks the right subset across studies.
3. Curation is a human judgment call. Run Selector is where that selection happens, and it exports the result as a one-column text file. Keeping the human in that loop beats any heuristic in ww-sra.

Additional context:

• This complements skip_on_chemistry_failure from Enabling "graceful fail" for CellRanger run_count tasks #357: since no curation guarantees a clean single-cell-only set, graceful fail stays as the net for a mis-picked accession (lands it on skipped_sample_list instead of failing the scatter).
• PR scope: ww-sra-cellranger.wdl (add sra_id_file, resolve sra_ids, update scatter + summarize_chemistry_status), parameter_meta, inputs.json, README.md. testrun.wdl stays zero-config on an inline list.
• Explicit non-goal: we are not adding a project-ID-to-run-list task to ww-sra.