pybullet_envs.deep_mimic: use getLinkStates for faster 'getReward'

pybullet_envs.deep_mimic: add profile timers for each thread to optimize training

examples/pybullet/gym/pybullet_envs/deep_mimic/DeepMimic_Optimizer.py

@@ -12,6 +12,7 @@ from pybullet_envs.deep_mimic.learning.rl_world import RLWorld
 from pybullet_utils.logger import Logger
 from pybullet_envs.deep_mimic.testrl import update_world, update_timestep, build_world
 import pybullet_utils.mpi_util as MPIUtil
+from mpi4py import MPI
 
 args = []
 world = None
@@ -20,9 +21,15 @@ world = None
 def run():
   global update_timestep
   global world
-
+  rank = MPI.COMM_WORLD.Get_rank()
+  name = "timings"+str(rank)+".json"
+  slot = world.env._pybullet_client.startStateLogging(world.env._pybullet_client.STATE_LOGGING_PROFILE_TIMINGS,name)
+  count=1000
   done = False
   while not (done):
+    count-=1
+    if (count<0):
+      world.env._pybullet_client.stopStateLogging(slot)
     update_world(world, update_timestep)
 
   return