Corrected MXNet's PPO Head for Continuous Action Spaces (#84)

* Changes required for Continuous PPO Head with MXNet. Used in MountainCarContinuous_ClippedPPO.

* Simplified changes for continuous ppo.

* Cleaned up to avoid duplicate code, and simplified covariance creation.

rl_coach/architectures/mxnet_components/architecture.py

@@ -299,8 +299,7 @@ class MxnetArchitecture(Architecture):
         assert outputs is None, "outputs must be None"
 
         output = self._predict(inputs)
-
-        output = tuple(o.asnumpy() for o in output)
+        output = list(o.asnumpy() for o in output)
         if squeeze_output:
             output = squeeze_list(output)
         return output