- use (starpu_cpu_func_t) 1 trick, same as cuda_func
    - cpu funtions are not defined anymore avoiding the dependency to coreblas
    - add #if !defined(CHAMELEON_SIMULATION) where it is needed
    - remove dependency to the coreblas library (become useless)
    - remove useless simucblas, simulapacke libraries
    - remove CHAMELEON_SIMULATION_MAGMA cmake variable and definition
      - keep using CHAMELEON_USE_CUDA and CHAMELEON_USE_MAGMA to consider CUDA kernels
      - this avoid to introduce useless new variables
    - work on messages

The new codelet is added for all 3 runtimes but only tested with starpu ;-)
An example test7 is added : it is a copy of test6 that uses the new function to build the matrices

Add CHAMELEON_SIMULATION_MAGMA option to select whether to enable magma kernels or not in simulation mode

does not build with starpu/simgrid if symbols starpu_mpi_cache_flush and starpu_mpi_cache_flush_all_data

Fix a large bunch of warnings, and there are still mistakes that should be fixed before the SC release

add cudablas library to make calls to cuda kernels (cublas here), no more calls to cublas in runtime/starpu codelets

add cudablas library to make calls to cuda kernels (magma here, cublas will follow), no more calls to magma in runtime/starpu codelets