Transfer between grids functionality for subgraph parallelism (elemen…

…tal#105)

* translate between grids support

* added tests and new unified function interface for subgraph parallelism

* code cleanup

* updated transferbetweengrid to support subgrids

* code cleanup

* Code cleanup

Co-authored-by: Tim Moon <[email protected]>
Co-authored-by: Tom Benson <[email protected]>

* replaced CudaDeviceSynchronize with El::gpu::SynchronizeDevice

* Added assertions for n evenly divided by size of parent grid and updated allreduce test

* Update include/El/blas_like/level1/Copy/TranslateBetweenGrids.hpp

* Update include/El/blas_like/level1/Copy/internal_decl.hpp

* Update include/El/blas_like/level1/Copy/TranslateBetweenGrids.hpp

Co-authored-by: Arpan Jain <[email protected]>
Co-authored-by: Arpan Nmn Jain <[email protected]>
Co-authored-by: Arpan Nmn Jain <[email protected]>
Co-authored-by: Arpan Jain <[email protected]>
Co-authored-by: Arpan Jain <[email protected]>
Co-authored-by: Tim Moon <[email protected]>
Co-authored-by: Tom Benson <[email protected]>
Co-authored-by: Arpan Jain <[email protected]>