Fixed-point matrix multiplication improvements #2791

Job	Run time
macOS-M1, Clang (x3)	8m 37s
GCC build regression check (no tests)	1m 4s
Alpine Linux, musl, 32-bit (assert, x1.5)	13m 59s
macOS-x86 Clang with BLAS (x2)	14m 58s
MinGW GCC (x0.5)	11m 3s
MSVC (x1)	25m 36s
Nemo.jl (temporary branch)	11m 57s
Ubuntu Clang with examples (x5)	12m 41s
Ubuntu GCC, Code Coverage (x10)	30m 4s
Ubuntu GCC with NTL (assert, x2)	8m 34s
	2h 18m 33s

Provide feedback