Update trainer config

config/sac_trainer_config.yaml

     memory_size: 128
     normalize: false
     num_update: 1
-    train_interval: 1
+    steps_per_update: 10
     num_layers: 2
     time_horizon: 64
     sequence_length: 64
     buffer_size: 500000
     max_steps: 2.0e6
     init_entcoef: 0.05
-    train_interval: 1
 
 Bouncer:
     normalize: true
     normalize: true
     time_horizon: 1000
     batch_size: 256
-    train_interval: 2
+    steps_per_update: 20
     buffer_size: 500000
     buffer_init_steps: 2000
     max_steps: 5e6
     batch_size: 256
     buffer_size: 500000
     summary_freq: 30000
-    train_interval: 2
+    steps_per_update: 20
     num_layers: 3
     max_steps: 1e7
     hidden_units: 512
     max_steps: 2e7
     summary_freq: 30000
     num_layers: 4
-    train_interval: 2
+    steps_per_update: 20
     hidden_units: 512
     reward_signals:
         extrinsic: