[Cell-level metadata] Implement in the Pipeline

[gerbeldo]

https://github.com/hms-dbmi-cellenics/pipeline/pull/328

Goal

• [x] Add cell-level-metadata bucket to bucket list
• [x] #59
• [x] Extract the cell_id <-> barcode map from the Seurat object
• [x] Download the cell-level metadata table from the S3 bucket, with the key received in the processing config and the bucket list (barcode <-> metadata map)
• [x] #60
• [x] Detect group variables using find_group_columns functions from the Seurat pipeline
• [x] Create cell_id <-> metadata map
• [x] Create corresponding cellsets (CLMPerSample for group vars and CLM for the others)
• [ ] Add to cellsets object and patch through the API (using existing functionality)