Add non-FFT based linear random wave model

[srmainwaring]

This PR adds a non-FFT variant of a linear random wave model for comparison and benchmarking.

Simulations for wave energy devices usually have some variant of a wave model comprised of a sum over linear waves with randomised phases. The sums are typically computed by simple loops over the components and the amplitudes are drawn from a suitable wave spectrum. While not efficient for simulating a wave field where samples must be taken at many points, the models is practical for simulating linear wave body interactions where the field is evaluated at the origin at each time step.

Details

• Superposition of num_waves waves.
• Sampled at constant angular frequency: max_omega / num_waves.
• Amplitudes determined by the Pierson-Moskowitz spectrum.
• Wave direction may be set.
• Waves do not spread, all waves propagate in the same direction.
• Waves are assigned random phases.
• Plots of wave elevation and pressure.

Other

• Minor formatting changes to LinearRegularWaveSimulation for consistency.

Usage

A limited set of parameters may be configured for the model and visual plugins.

<plugin
    filename="gz-waves1-waves-model-system"
    name="gz::sim::systems::WavesModel">
  <static>0</static>
  <update_rate>30</update_rate>
  <wave>
    <algorithm>linear_random</algorithm>
    <tile_size>256</tile_size>
    <cell_count>128</cell_count>
    <wind_speed>5.0</wind_speed>
    <wind_angle_deg>135</wind_angle_deg>
  </wave>
</plugin>

<plugin
    filename="gz-waves1-waves-visual-system"
    name="gz::sim::systems::WavesVisual">
  <static>0</static>
  <mesh_deformation_method>DYNAMIC_GEOMETRY</mesh_deformation_method>
  <tiles_x>-0 0</tiles_x>
  <tiles_y>-0 0</tiles_y>
  <wave>
    <algorithm>linear_random</algorithm>
    <tile_size>256</tile_size>
    <cell_count>128</cell_count>
    <wind_speed>5.0</wind_speed>
    <wind_angle_deg>135</wind_angle_deg>
  </wave>
</plugin>

Future work

• Provide configuration for remaining parameters (num waves, frequency bounds, spectrum choice).
• Provide more comprehensive test suite - check realised elevation variance vs. expectation from spectrum.
• Improve visuals (appears to be some lighting artefacts - possibly tangent space calc).
• Improve tiling - same issue as simple regular wave models - the tile skirt does not appear to be treated correctly.