[BUG] fix Mixture distribution for more than two components (#337)

The `Mixture` distribution after API upgrade broke for 3 or more
components.

This has been fixed, and is now covered by a test case.

skpro/distributions/mixture.py

@@ -225,7 +225,8 @@ def sample(self, n_samples=None):
         dists = [d[1] for d in self._distributions]
         raw_samples = [d.sample(N).values for d in dists]
         masked_samples = [ind * raw for ind, raw in zip(indicators, raw_samples)]
-        sample = np.add(*masked_samples)
+        masked_samples = np.array(masked_samples)
+        sample = masked_samples.sum(axis=0)
 
         if n_samples is None:
             spl_index = self.index
@@ -273,12 +274,18 @@ def get_test_params(cls, parameter_set="default"):
         params2 = {"distributions": dists2, "weights": [0.3, 0.7]}
 
         # scalar case
-        normal1 = Normal(mu=0, sigma=1)
-        normal2 = Normal(mu=3, sigma=2)
-        dists = [("normal1", normal1), ("normal2", normal2)]
-        dists2 = [normal1, normal2]
+        normal3 = Normal(mu=0, sigma=1)
+        normal4 = Normal(mu=3, sigma=2)
+        dists = [("normal3", normal3), ("normal4", normal4)]
+        dists2 = [normal3, normal4]
 
         params3 = {"distributions": dists2}
         params4 = {"distributions": dists, "weights": [0.3, 0.7]}
 
-        return [params1, params2, params3, params4]
+        # more than 2 distributions
+        normal5 = Normal(mu=[[0, 1], [2, 3], [4, 5]], sigma=2, columns=columns)
+        normal6 = Normal(mu=[[0, 1], [2, 3], [4, 5]], sigma=0.5, columns=columns)
+        dists3 = [normal1, normal2, normal5, normal6]
+        params5 = {"distributions": dists3}
+
+        return [params1, params2, params3, params4, params5]