Use clang instead of gcc and compile with parallel threads when generating FMUs

[casella]

As discussed in #11759, when we generate FMUs under Windows (not sure about Linux) we still use gcc and compile the files one at a time, which makes the whole thing horribly slow.

We should use clang and parallel processes instead, as we do for regular simulations.

[casella]

Preferably use the same settings that we have for regular simulations in OMEdit

[AnHeuermann]

For general changes that should be applied to all compilations with CMake add options to OMCompiler/SimulationRuntime/fmi/export/buildproject/CMakeLists.txt.in.

If you want to add something that should depend on some omc setting easiest way is to add it in CevalScriptBackend.mo, like it was done for CMAKE_BUILD_TYPE.

Change compiler: cmake -DCMAKE_CXX_COMPILER=/pathto/clang++ -DCMAKE_C_COMPILER=/pathto/clang [...]
Number of threads: cmake --build . --target -j<N>

[casella]

Adding @deepak19015 to the loop

[casella]

Adding @Michelecroce to the loop

[adrpo]

I guess this is now fixed via #11904. Reopen if there are any issues.

[casella]

Thanks @adrpo, I'll test this ASAP. If it works fine, I would port it to maintenance/v1.22, so I'm reopen it as a reminder

[casella]

I just generated an FMU of Buildings.Examples.ChillerPlant.DataCenterContinuousTimeControl using the latest nightly build on my Windows PC, it took about one minute (which is reasonable) and I checked with the Process Monitor that the compilation process was indeed using clang in parallel. Success!

@adrpo please cherry-pick this to maintenance/v1.22 and close the ticket.

[casella]

Merged into maintenance/v1.22 in #11923