chore: lower compute resource requirements for dataset metadata update

[nayib-jose-gloria]

Reason for Change

• lower resources required for dataset metadata update batch job

Changes

• use h5py.File to edit uns on h5ad file without reading in or re-writing out dfs or matrices
• remove rpy2 dependency from processing image now that seurat is deprecated

Testing steps

• rdev testing [tbd]

[github-actions]

Deployment Summary

• backend_url: https://pr-7413-backend.rdev.single-cell.czi.technology
• backend_de_url: https://pr-7413-backend-de.rdev.single-cell.czi.technology
• backend_wmg_url: https://pr-7413-backend-wmg.rdev.single-cell.czi.technology
• frontend_url: https://pr-7413-frontend.rdev.single-cell.czi.technology
• delete_db_task_definition_arn: arn:aws:ecs:us-west-2:***:task-definition/dp-rdev-pr-7413-deletion:1
• migrate_db_task_definition_arn: arn:aws:ecs:us-west-2:***:task-definition/dp-rdev-pr-7413-migration:1