Fix release 0.8.5 tests for PPO torch Breakout. (#8226)

ci/regression_test/rllib_regresssion_tests/compact-regression-test.yaml

@@ -52,7 +52,7 @@ atari-ppo-torch:
     stop:
         time_total_s: 3600
     config:
-        use_pytorch: true,
+        use_pytorch: true
         lambda: 0.95
         kl_coeff: 0.5
         clip_rewards: True

rllib/agents/ppo/ppo.py

@@ -179,7 +179,7 @@ def validate_config(config):
 
 
 def get_policy_class(config):
-    if config.get("use_pytorch") is True:
+    if config["use_pytorch"]:
         from ray.rllib.agents.ppo.ppo_torch_policy import PPOTorchPolicy
         return PPOTorchPolicy
     else: