Revert adding weighting to curve fitting.

This code was never meant to be added to the mainline, it doesn't do
anything, and actually causes problems if it is used.

Revert to a point before commit
a30ed72.

lib/rust/mmscenegraph/src/math/curve_fit.rs

@@ -394,8 +394,6 @@ pub struct CurveFitLinearNPointsProblem {
     reference_values: Vec<(f64, f64)>,
     // X-coordinates of control points (fixed).
     control_points_x: Vec<f64>,
-    // Weights.
-    weights: Vec<f64>,
     // Interpolation method to use.
     interpolator: Interpolator,
 }
@@ -404,21 +402,14 @@ impl CurveFitLinearNPointsProblem {
     fn new(
         control_points_x: Vec<f64>,
         reference_curve: &[(f64, f64)],
-        weights: &[f64],
         interpolation_method: InterpolationMethod,
     ) -> Self {
         let reference_values: Vec<(f64, f64)> =
             reference_curve.iter().copied().collect();
 
-        let weights_max =
-            weights.iter().fold(Real::NEG_INFINITY, |a, &b| a.max(b));
-        let weights: Vec<f64> =
-            weights.iter().map(|w| w / weights_max).collect();
-
         Self {
             reference_values,
             control_points_x,
-            weights,
             interpolator: Interpolator::from_method(interpolation_method),
         }
     }
@@ -440,15 +431,12 @@ impl CurveFitLinearNPointsProblem {
     }
 
     fn residuals(&self, control_points_y: &[f64]) -> Vec<f64> {
-        self.weights
+        self.reference_values
             .iter()
-            .zip(self.reference_values.iter())
-            .map(|(weight, &(value_x, data_y))| {
+            .map(|&(value_x, data_y)| {
                 let curve_y =
                     self.interpolate_y_value_at_x(value_x, control_points_y);
-                // TODO: Can we use the 'weight' for each point or is
-                // that invalid?
-                (curve_y - data_y).abs() // * weight
+                (curve_y - data_y).abs()
             })
             .collect()
     }
@@ -581,9 +569,6 @@ pub fn nonlinear_line_n_points_with_initial(
     y_values: &[f64],
     x_initial_control_points: &[f64],
     y_initial_control_points: &[f64],
-    // Per-value weights, so we can give more emphasis to more
-    // important frames, and we can essentially filter out 'bad' data.
-    weights: &[f64],
     interpolation_method: InterpolationMethod,
 ) -> Result<Vec<Point2>> {
     assert!(x_values.len() > 2);
@@ -601,7 +586,6 @@ pub fn nonlinear_line_n_points_with_initial(
         y_initial_control_points.len(),
         "X and Y control point values must match length."
     );
-    assert_eq!(x_values.len(), weights.len());
     let control_point_count = x_initial_control_points.len();
 
     // Create reference values.
@@ -615,7 +599,6 @@ pub fn nonlinear_line_n_points_with_initial(
     let problem = CurveFitLinearNPointsProblem::new(
         x_initial_control_points.to_vec(),
         &reference_values,
-        &weights,
         interpolation_method,
     );
 
@@ -672,8 +655,6 @@ pub fn nonlinear_line_n_points(
         "Must have at least 3 control points"
     );
 
-    let weights = vec![0.0; x_values.len()];
-
     let mut point_x = 0.0;
     let mut point_y = 0.0;
     let mut dir_x = 0.0;
@@ -702,7 +683,6 @@ pub fn nonlinear_line_n_points(
         y_values,
         &x_initial_control_points,
         &y_initial_control_points,
-        &weights,
         interpolation,
     )
 }