Bench cpp

[imrn99]

Added:

• CPP counterparts of the BLAS kernels benchmarks, written using Kokkos. Basic cmake setup is also provided.
• Layout effect benchmarks, testing worst case, usual case, and best case layout for matrix-matrix product.
• Data size effect on layout choice benchmarks, testing usual case and best case layout over a range of data sizes for matrix-matrix product.

Also changed the config of the release profile to have slightly better perf. Overall improvement might be negligible.