Abstract vessel performance into configs, generalise performance calculation

[thomaszwagerman]

Closes #336, these are very much breaking changes. Should be released as 2.0.0.

Target will be main once tz/ruff -> v1.1.9 are merged.

Added

• Model Registry System: Plugin-based architecture for resistance and consumption models using @register_model decorator.
• Generic Vessel Classes: Ship, Glider, AUV, and Aircraft classes replacing hard-coded vessel types.
• Resistance Models: Three pluggable resistance models:
  • ice_froude: Froude number-based ice resistance with configurable k/b/n coefficients
  • wind_drag: Angle-dependent wind drag with interpolation options (linear/cubic/spline)
  • wave_kreitner: Kreitner wave resistance with beam/length/block coefficient parameters
• Consumption Models: Three pluggable consumption models:
  • polynomial_fuel: Speed and resistance-based fuel consumption with polynomial coefficients
  • polynomial_battery: Speed and depth-based battery consumption for underwater vehicles
  • constant_consumption: Fixed-rate consumption for simple vehicles
• Configuration Validation: Comprehensive JSON schema for vessel configurations with detailed parameter documentation including units.
• Documentation:
  • Vessel configuration gallery with ready-to-use examples (docs/config/vessel_gallery.md)
  • Resistance models theory documentation (docs/methods/resistance_models.md)
  • Complete model parameters reference (docs/reference/model_parameters.md)
• Example Configurations: Updated vessel configs for SDA, SDA_wind, Slocum, and BoatyMcBoatFace using new format.
• Unit Tests: 27 comprehensive tests covering model registry, all resistance/consumption models, generic vessel classes, and factory.

Changed

• Vessel Configuration Format: vessel_type replaced with vessel_class + explicit model specifications.
• Configuration Structure: All vessel-specific parameters (beam, force_limit, etc.) now explicitly defined in model params rather than top-level.
• VesselFactory: Simplified factory using VESSEL_CLASSES mapping instead of hard-coded requirements dictionary.
• Physical Constants: Now configurable per-vessel (gravity, air_density, rho_water) with sensible defaults.
• Model Composition: Vessels compose multiple resistance models; total resistance is sum of all active models.

Removed

• Legacy Vessel Classes (BREAKING): Deleted 12 hard-coded vessel files:
  • SDA.py, SDA_wind.py, example_ship.py (ships)
  • slocum.py (gliders)
  • boatymcboatface.py (AUVs)
  • twin_otter.py, windracer.py (aircraft)
  • abstract_ship.py, abstract_glider.py, abstract_alr.py, abstract_plane.py, abstract_uav.py (abstract classes)
• Configuration Fields (BREAKING):
  • vessel_type (replaced by vessel_class)
  • hull_type (slender/blunt distinction now encoded in k/b/n coefficients)
  • Top-level beam, force_limit (now in resistance model params)

[thomaszwagerman]

Marking some specific vessel performance tests as XFAIL, see #344

[thomaszwagerman]

Decision should be made whether we rename any instance of vessel to vehicle. I think given the inclusion of aircraft and future development work, the earlier we do this the better.

This branch already includes breaking changes, but this would introduce another significant breaking change.

[thomaszwagerman]

Just verifying that:

SDA_VESSEL_CONFIG_MAIN = {
    "vessel_type": "SDA",
    "max_speed": 26.5,
    "unit": "km/hr",
    "beam": 24.0,
    "hull_type": "slender",
    "force_limit": 96634.5,
    "max_ice_conc": 80,
    "min_depth": 10,
}

on main, and the new config on tz/vessel_performance

SDA_VESSEL_CONFIG_NEW = {
    "vessel_class": "ship",
    "max_speed": 26.5,
    "unit": "km/hr",
    "max_ice_conc": 80,
    "min_depth": 10,
    "num_directions": 8,
    "resistance_models": [
        {
            "type": "ice_froude",
            "params": {
                "k": 4.4,
                "b": -0.8267,
                "n": 2.0,
                "beam": 24.0,
                "force_limit": 96634.5,
                "gravity": 9.81,
            },
        },
        {
            "type": "wind_drag",
            "params": {
                "frontal_area": 750.0,
                "air_density": 1.225,
                "interpolation": "linear",
                "angles": [0, 30, 60, 90, 120, 150, 180],
                "coefficients": [0.94, 0.77, 0.42, 0.48, 0.17, -0.30, -0.34],
            },
        },
    ],
    "consumption_model": {
        "type": "polynomial_fuel",
        "params": {
            "speed_coeffs": [0.00137247, -0.0029601, 0.25290433],
            "resistance_coeffs": [7.75218178e-11, 6.48113363e-06],
        },
    },
}

produce identical results on the scalar_grf case:

Some better regression testing would go a long way here.