Persistent PoolManager request queue

[kofemann]

The request container in the pool manager uses an in-memory store to keep track of active stage and p2p requests. As a result, on the restart, PoolManager 'forgets' all ongoing stages and will trigger a new stage if a new request for a file that is already being staged. Moreover, if the ongoing stage competes, PoolManager is not notified about the file being online and will wait until the newly started stage request is completed.

A possible solution to this issue can be:

• persistent store for ongoing stage request
• check file availability when PoolManager receives a new request