HDF5 Build Update, 1.0 branch (2018.03.16.)

While running my final tests with the LCG_93 updates, I had to realise that the way HDF5 was compiled for LCG_93 does not work with our current build setup.

You see, CMake's FindHDF5.cmake module is very determined to make use of the h5cc and h5c++ scripts, if they exist. They both do in LCG_93. So CMake could just not be convinced to pick up HDF5 from our own external projects instead.

Since these two scripts are only built when using the autoconf/automake build of HDF5, I now reluctantly switched our build to work like that. Taking the LCG_93 configuration as reference.

At this point I still had to tweak our FindHDF5.cmake module a little to make Athena based projects find the right version of HDF5 during the build.

But with all of this done, I believe this configuration should be a lot more robust than what we had before. @dguest, please have a look, and tell me (quickly) if you disagree with these updates for some reason. Since without them I won't be able to move ahead with the LCG_93 update of the 21.2 branch.