Improve argument parsing in tests

[pbartholomew08]

          I feel like this is a design issue. Perhaps we could have a single init subroutine that accepts an array of arguments and the calling code maps these into 2decomp values?

Originally posted by @pbartholomew08 in https://github.com/2decomp-fft/2decomp-fft/pull/350#discussion_r1773097392

[mathrack]

In my opinion, a proper design could be to use command-line arguments such as

mpirun -n 5 test --nx 32 --ny 64 --nz 128 --prow 1 --pcol 5 ....

However I am not sure it can be done easily without introducing extra dependencies.