Ready Vectorization API for Deployment

[rchen20]

To Do:

• Fix tensor matrix col-major subtract tests fail #1485
• Check these:
  • Vectorization Compilation Problem on x86-Cuda Builds #1581
  • Issue building on Perlmutter. #1464
  • RAJA Build fails in avx2_int32.hpp when using Cray MPI Compiler wrappers #1296
  • Issue with RAJA header file while compiling on Fedora 35 with gcc-11.2.1 #1351
• Look in to adding ARM vector intrinsics.
• Solve the performance issue with loads and stores being interspersed with computation.
• Carve out matrix interface, and move the vector-register API out of expt namespace.