Cold Storage: Update S3 storage class instead of moving objects to a dedicated bucket

Today, the implementation to send a blob to cold storage relies upon the Blob Dispatcher.

When a blob has the ColdStorage facet, a blob dispatcher rule moves the main blob from the default bucket to another bucket that must be configured to set the proper cold storage class (glacier lifecycle rule for AWS s3) on the stored blobs.

We'd like to evolve this paradigm by letting the blobs at their original location (bucket) and explicitly update their storage class:

• in case of a mass movement to cold storage (through BAF NXP-30948), moving Terabytes of blobs to another bucket won't scale
• We don't have an auto GC that would clean up the blob from the default bucket and we end up with more storage while we try to reduce the cost of the storage (NXP-30949)
• isolating blob sent to Cold storage in a dedicated bucket prevents from setting them under retention because documents under retention are also sent to another dedicated bucket. If we can just update the storage class wherever the blob are located, it would leave the door open to both retention and cold storage being applied to a same document/blob