Missing compilation flags using spec_wrapper

[missing-user]

I noticed, that SPEC was significantly slower when running from simsopt than standalone. Looking closer, the shared object Library produced when compiling spec_wrapper doesn't utilize the same compilation flags, e. g. -O3.

When running the initialization from the spec wrapper, the console output also indicates no compilation flags were passed.

       :            : fc      = /usr/bin/mpifort ; 
       :            : flags   =  ; 

I think we should fix spec_wrapper/CmakeLists.txt to at least enable ftree-vectorize, such that xspec standalone and the Python bindings behave the same.

[missing-user]

The question was promted because I noticed that some parts of SPEC were running on one thread, eventho they should have been OMP parallel (e.g. poincare, virtual casing)
After a discussion with @smiet it seems like this is specific to GCC and works fine with the Intel compiler. The missing compilation flags should be added to the call to f2py in the future.