Reduced the size of the forces array that is manipulated in matrix ve… #2996

Job	Run time
conda / conda (ubuntu-20.04)	11m 10s
docker / docker (fedora39)	19m 44s
codecheck / codecheck	43m 41s
docker / docker (rocky8)	23m 12s
linux / linux (-doc-mpi-)	38m 37s
conda / conda (macos-12)	18m 34s
linux / linux (-mpi-)	49m 55s
conda / conda (macos-14)	7m 16s
linux / linux (-coverage-mpi-)	1h 1m 3s
mac / macports (macos-12)	16m 33s
mac / macsimple (macos-12)	9m 12s
linux / linux (-debug-)	28m 2s
mac / macports (macos-14)	8m 45s
linux / linux (-debug-mpi-)	38m 57s
mac / macports (+allmodules, macos-12)	19m 43s
mac / macsimple (macos-14)	7m 10s
linux / linux (-intel-)	47m 15s
linux / linux (-pycv-mpi-)	18m 24s
	7h 47m 13s

Provide feedback