Make solve_ΩplusK MPI-compatible

[Technici4n]

Only the dot and norm in the CG solver had to be made MPI-compatible.

Some quick numbers from benchmarking an operation dominated by solve_ΩplusK on my computer:

MPI processes	running time
1	97.7s
2	49.4s
4	27.5s

[antoine-levitt]

Nice! Can you move mpi_dot and mpi_norm to mpi.jl? (potentially we should audit the code base to see if there are other places that could use it as well) Also not worth documenting the mpi in that file (things are already complicated enough and the mpi is pretty simple here).

[Technici4n]

Done.

potentially we should audit the code base to see if there are other places that could use it as well

I just did a quick search for mpi_sum and did not find other places :)