Merge branch 'improve_and_bugfix_low_and_high_level_training' into 'master'

backends/controller_base.py

backends/kerasrl_learner.py

backends/manual_policy.py

.../trained_policies/0.1mil_weights/follow_weights_actor.h5f

...trained_policies/0.1mil_weights/follow_weights_critic.h5f

...rained_policies/0.1mil_weights/keeplane_weights_actor.h5f

...ained_policies/0.1mil_weights/keeplane_weights_critic.h5f

.../trained_policies/changelane/changelane_weights_actor.h5f

...trained_policies/changelane/changelane_weights_critic.h5f

backends/trained_policies/follow/follow_weights_actor.h5f

backends/trained_policies/follow/follow_weights_critic.h5f

...nds/trained_policies/halt/immediatestop_weights_actor.h5f

...ds/trained_policies/halt/immediatestop_weights_critic.h5f

backends/trained_policies/highlevel/highlevel_weights.h5f

...ends/trained_policies/highlevel/highlevel_weights_772.h5f

...ends/trained_policies/keeplane/keeplane_weights_actor.h5f

...nds/trained_policies/keeplane/keeplane_weights_critic.h5f

backends/trained_policies/left/left_weights_actor.h5f

backends/trained_policies/left/left_weights_critic.h5f

backends/trained_policies/right/right_weights_actor.h5f