Put cling pragma for loading fftw3 in headers

[egpbos]

As in https://github.com/QuantStack/xtensor-io/blob/a57e3b90ca1025a180c5813c56b8a344ec2d8165/include/xtensor-io/xaudio.hpp#L20-L21, we can put the library loading step in the headers themselves, which saves cling/notebook users from doing it.

[egpbos]

Also put this in CMakeLists.txt and in xtensor-fftw_config.hpp:

https://github.com/QuantStack/xtensor-io/blob/a57e3b90ca1025a180c5813c56b8a344ec2d8165/CMakeLists.txt#L18

This makes sure the library paths get properly added. Right now, the notebooks are already working, because I'm loading xtensor-io, which also sets all this stuff. Without xtensor-io and without the cling pragma, fftw calls would fail.