Call Constraints

Note the following sequence when calling thread functions:

1. Firstly, call kml_fft(f/h)_init_threads to initialize thread use.
2. Secondly, call kml_fft(f/h)_plan_with_nthreads to specify the number of threads.
3. Lastly, call kml_fft(f/h)_cleanup_threads to release resources related to the multi-thread framework.