Update kalmanfilter.py for lower memory usage (#26)

---------

Co-authored-by: Colin Catlin <colin.catlin@gmail.com>

simdkalman/kalmanfilter.py

@@ -107,11 +107,14 @@ def __init__(self, mean, cov):
         self.mean = mean
         if cov is not None:
             self.cov = cov
+        else:
+            self.cov = None
 
     @staticmethod
     def empty(n_states, n_vars, n_measurements, cov=True):
         mean = np.empty((n_vars, n_measurements, n_states))
         if cov:
+            # for lower memory switch to ...n_states), dtype=np.float32)
             cov = np.empty((n_vars, n_measurements, n_states, n_states))
         else:
             cov = None
@@ -383,7 +386,8 @@ def smooth(self,
         :type states: boolean
         :param observations: return smoothed observations :math:`y`?
         :type observations: boolean
-        :param covariances: include covariances results?
+        :param covariances: include covariances results? For lower memory
+            usage, set this flag to ``False``.
         :type covariances: boolean
 
         :rtype: Result object with fields
@@ -557,15 +561,15 @@ def auto_flat_states(obs_gaussian):
                 result.smoothed.states = empty_gaussian()
 
                 # lazy trick to keep last filtered = last smoothed
-                result.smoothed.states.mean = 1*filtered_states.mean
+                result.smoothed.states.mean = filtered_states.mean
                 if covariances:
-                    result.smoothed.states.cov = 1*filtered_states.cov
+                    result.smoothed.states.cov = filtered_states.cov
 
             if observations:
                 result.smoothed.observations = empty_gaussian(n_states=n_obs)
-                result.smoothed.observations.mean = 1*filtered_observations.mean
+                result.smoothed.observations.mean = filtered_observations.mean
                 if covariances:
-                    result.smoothed.observations.cov = 1*filtered_observations.cov
+                    result.smoothed.observations.cov = filtered_observations.cov
 
             if gains:
                 result.smoothed.gains = np.zeros((n_vars, n_measurements, n_states, n_states))