Remove check of y being a subset of X in AnomalyView

It doesn't need to be, as this logic is up to the .anomaly()
implementation of the model being served. Currently, DiffBasedAnomaly
does not require y to be a subset of X, because it only compares
y and yhat.

gordo_components/server/views/anomaly.py

@@ -145,12 +145,6 @@ def _create_anomaly_response(self, start_time: float = None):
             }
             return make_response((jsonify(message), 400))
 
-        # It is ok for y to be a subset of features in X, but we need at least one
-        # to compare against to calculate an error.
-        if not any(col in g.y.columns for col in [t.name for t in self.tags]):
-            message = {"message": "y is not a subset of X, cannot do anomaly detection"}
-            return make_response(jsonify(message), 400)
-
         # Now create an anomaly dataframe from the base response dataframe
         try:
             anomaly_df = current_app.model.anomaly(g.X, g.y, frequency=self.frequency)