feat(cellranger/count): support SRA three-file FASTQ inputs (_1/_2/_3) in renaming logic

## Summary This PR extends the `cellranger count` module’s FASTQ renaming logic to handle SRA / `fasterq-dump` outputs that produce three files per sample (`*_1/_2/_3.fastq(.gz)`), which is common for 10x chemistries when run with `--split-files --include-technical`. Specifically: - Detect SRA-style suffixes `_1`, `_2`, `_3` in filenames. - Group files by prefix and, when `_2` and `_3` are present, map: - `_2` → `R1` (barcode+UMI), - `_3` → `R2` (cDNA), - `_1` (if present) → `I1` (sample index). - Keep existing behaviour for bcl2fastq-style `R1/R2` filenames unchanged. - Emit a clear error if neither pattern (bcl2fastq or SRA `_1/_2/_3`) can be recognised. No changes were required in `modules/nf-core/cellranger/count/main.nf`; all logic is contained within `cellranger_count.py`. ## Motivation / use case Many public 10x datasets are only available in SRA. When using: ```bash prefetch SRR9304758 fasterq-dump --split-files --include-technical SRR9304758.sra