Wisdem

The PR brings in several changes:
Blade structure can be parametrized in the thickness of each layer, not only spar caps and reinforcements
There are new blade constraints to support inverse design studies (desired blade mass, desired natural frequencies)
New example and documentation to support inverse design
Update of installation steps (mamba)
Turbine gearbox can be desired with a desired torque density
User specified DVs, figure of merit, and constraints in support of WEIS Phase II
Strengthening of the sub-optimization to determine pitch angle across wind speeds in rotor_power.py
Cd0 no longer hard coded to 0
The list probably goes on

Highlights include:

- Better ability to adjust thickness as a design variable
- Changes to support the new IEA Wind 22-MW Reference Wind Turbine
- Better handling of multiple airfoil polars in the yaml-input file
- Remove duplicate code and capability between CCBlade and AirfoilPrep (AirfoilPrep has been fully removed)
- Fix bug in blade cross-sectional area calculation. Instead of doing this in PreComp, it is now done as a layup summation.
- Fix bug in sizing of main bearings for select drivetrain configurations

The next development tasks will involve more significant refactoring of the code to allow user-definitions of optimization problems and component mass / cost / stiffness values!

This is a minor release that includes:

- Careful and methodical improvements to the optimization approach, robustness, and results in FloatingSE. We still strongly recommend floating platform design with [WEIS](https://github.com/WISDEM/WEIS) (in the frequency or time domain), but FloatingSE is improving as a starting point thanks to the work of dzalkind
- Bug fix so that the hub and nacelle center of gravity and moment of inertia calculation accounts for blade cone angle
- Improved documentation and units consistency in the NREL Cost and Scaling Model
- A handful of other bug fixes

This is a minor release PR with a number of small fixes:

- Fix issues with upstream library changes in openmdao, pytest, and numpy that made life difficult for users and the CI builds
- Remove moorpy from the codebase as we can now rely on its pypi or conda versions
- Output labor hours required for towers, monopiles, and jackets
- Relax a hard limits on peak thrust shaving for a smoother thrust profile
- Fixes in the yaml input schema