RustFS — removed

The third-party RustFS product (rustfs/rustfs) is no longer a first-party Atlas backend.

What changed

• Crate atlas-driver-rustfs deleted.
• Console Storage → RustFS, admin proxy, drive/instance jobs, lab chart, and deploy/rustfs-lab removed.
• POST /buckets defaults to Ceph RGW (bkd_ceph_lab) via Rook ObjectBucketClaim.
• backend_id: "bkd_rustfs_lab" is rejected.
• Persisted JobSpec variants bucket.create.rustfs / bucket.delete.rustfs / rustfs.drive.provision / rustfs.instance still deserialize so old job rows do not poison the queue; dispatch returns a retirement error.

What to use instead

Need	Use
New buckets	omit backend_id or pass "bkd_ceph_lab"
Presigned upload/download	existing /buckets/{id}/objects/* on RGW
BYO S3 (MinIO, Garage, customer RustFS, AWS)	atlas-driver-rgw::S3Target + endpoint/credentials Secret
DataBridge object copy	external S3 endpoint or Ceph RGW backend id
State backup	ATLAS_S3_ENDPOINT / ATLAS_RGW_PUBLIC_ENDPOINT

BackendType::Rustfs remains on the wire so existing inventory rows deserialize. Do not register new backends of that type.

ATLAS_RUSTFS_* environment variables are ignored (a warning is logged if ATLAS_RUSTFS_ENABLE=1).