Remove tile life from remaining routines

[neil-lindquist]

This PR removes tile life from the remaining routines. This doesn't do the big refactor to actually remove the tile life infrastructure, so that the review can focus on the logic for memory management.

A notable change is that I extended the releaseRemoteWorkspace functions to reduce multiple receive counts simultaneously, so I can get a little extra parallelism out of hbmm.

[ayarkhan]

Question: In the newly added release tasks you sometimes use in and sometimes inout for dependencies. Given that you are usually changing (freeing) data, shouldn't inout be the norm?

#pragma omp task depend(in:row[k])
#pragma omp task depend(inout:row[k])

[neil-lindquist]

The cases I used in were routines where we used backward looking dependency (i.e., gemm[k-1]) in the compute routines. In those cases, using inout would put the release task on the critical path.