Bug fix: revise verify_convergence to support multiple wind speeds

[Bartdoekemeijer]

Feature or improvement description

This feature fixes bug #383 where verify_convergence does not work if more than one wind speed is defined in FLORIS.

Related issue, if one exists

Issue #383

Impacted areas of the software

Yaw optimization.

Additional supporting information

In the past, we would first locate which turbines have a nonzero yaw angle and at what wind directions and wind speeds. We would only check the wind farm's power production for those turbines and conditions with and without yaw misalignment. With the grid-style way FLORIS is calculated, it's not possible to isolate exactly which wind direction/speed combinations we need to evaluate without doing a one-by-one loop, which is slow. Hence, the current method is simplified and just evaluates all atmospheric conditions num_turbs times, and for each atmospheric condition set we reset one of the turbine's yaw angle to its baseline value and calculate the power production. This means we may do many more calculations than necessary, but it does keep things in gridded format.

Test results, if applicable

Running example 09_opt_yaw_multiple_ws.py with verify_convergence=False yields:

In which we see nonzero yaw angles at high wind speeds of 15-17 m/s, due to numerical precision. These yaw angles just shift the cp value ever so slightly that the wind farm power is slightly higher. With verify_convergence=True, it yields:

Though, I have seen that the computation time can increase by 10-20% using verify_convergence=True. Hence, it still defaults to False and it's up to the user to decide whether one would like to include this functionality. Perhaps once the timeseries calculation functionality for FLORIS is implemented (#299) I can restructure this code to more efficiently evaluate a minimal set of conditions.

[Bartdoekemeijer]

Also fixed the example case and added explanation on how verify_convergence should be used. Once @paulf81 approves, I think this is ready to be merged in.

[paulf81]

This looks good to me @Bartdoekemeijer , I tested example 09 and looks great