This ticket is to implement the actual contents of the WT_SESSION::flush_tier API.

Things to think about:

• How do we return the object cookie to the caller? Possibly via a callback because every table will have its own address cookie as there could be thousands of them.
• Does flush_tier need to be atomic? As in, are metadata changes tracked on a per-table level or per-flush_tier level? (Consideration similar to a full database checkpoint.) It probably has to be at whatever level the flush to cloud storage is done.

Although this is discussed in the design along with block manager and will use the block manager API, it is a sizable chunk of code to be split into its own ticket.