Code cleanup (#1)

* cleaning up IMPALA and commenting DQN and DDPG scripts

* collector_device amend

* _make_collector helper with num_env_per_collector=1

* _make_collector helper with num_env_per_collector=1, bugfix

* annealing in terms of frames only

* update pbar

* passing device to collector (passing and policy)

* default pin_memory to False

* num_workers -> env_per_collector in ddpg make_parallel_env

* inplace bug in ddpg loss

* detach in ddpg loss

* _forward_hook_safe_action set_ bugfix

* contiguous sampling

* removing timing strings from ddpg

* PPO flags helpers

* PPO flags helpers, bugfix

* PPO flags helpers, bugfix

* PPO flags helpers, bugfix

* PPO flags helpers, bugfix

Author: vmoens