Implement Eq, Ord, and Hash for Scalar

These implementations aren't available for `f64` and other
floating-point values. Providing them is the whole point of why `Scalar`
exists.

src/math/scalar.rs

@@ -1,4 +1,4 @@
-use std::{cmp, f64::consts::PI, ops};
+use std::{cmp, f64::consts::PI, hash::Hash, ops};
 
 use approx::AbsDiffEq;
 
@@ -84,6 +84,22 @@ impl Scalar {
     }
 }
 
+impl Eq for Scalar {}
+
+impl Ord for Scalar {
+    fn cmp(&self, other: &Self) -> cmp::Ordering {
+        // Should never panic, as `from_f64` checks that the wrapped value is
+        // finite.
+        self.partial_cmp(&other).unwrap()
+    }
+}
+
+impl Hash for Scalar {
+    fn hash<H: std::hash::Hasher>(&self, state: &mut H) {
+        self.0.to_bits().hash(state);
+    }
+}
+
 impl From<f32> for Scalar {
     fn from(scalar: f32) -> Self {
         Self::from_f64(scalar as f64)