gp math overhaul to support more diverse tensor shapes #464

Job	Run time
build-numpy (3.8, short)	45s
build-docs (3.8)	3m 35s
build-mpi (3.9)	6h 0m 16s
build-numpy (3.8, optimize)	45s
build-jax (3.8)	1m 21s
build-mpi (3.10)	1m 38s
build-numpy (3.8, multivariate)	47s
build-torch (3.8)	2m 5s
build-jax (3.9)	57s
build-numpy (3.8, optimize-experimental)	47s
build-torch (3.9)	2m 16s
build-jax (3.10)	55s
build-numpy (3.9, short)	56s
build-torch (3.10)	2m 16s
build-numpy (3.9, optimize)	48s
build-numpy (3.9, multivariate)	50s
build-numpy (3.9, optimize-experimental)	50s
build-numpy (3.10, short)	49s
build-numpy (3.10, optimize)	49s
build-numpy (3.10, multivariate)	53s
build-numpy (3.10, optimize-experimental)	55s
	6h 25m 13s

Provide feedback