In WT-7844 the first issue found by the tiered_abort test is a race where, after a flush_tier call has completed its switching, and is waiting for the internal thread to finish copying the objects, a checkpoint starts. That checkpoint is looking for older objects in the bucket and they are not yet there. See that ticket for more information.

This ticket is to fix that issue. One potential solution is to have the block manager opener (or something else) look in the local system first, then the shared storage cache and then finally the shared storage bucket.