Implement PERSIST within a graph

Currrently ": PERSIST" creates a separate workflow action, which means that any activities shared between two persists may be duplicated if they are not also a workflow action.

Another alternative would be to generate PERSISTS within graphs. Updates would be similar to the ,UPDATE option we already have on disk writes. The code generator would also need to insert NOFOLDs on the boundaries between this persist and any others to ensure optimizations didn't percolate into the persists - causing the code hashes to change.

The syntax could possibly be different e.g., PERSIST(dataset, 'name).

There are issues of persist locking, and Thor being blocked from executing because it is waiting for a persist.