Re-timing post-processing improves motion generation quality at slow speeds, API documentation

New Features

• Significant improvements for generating slow trajectories. Added re-timing post processing to
  slow down optimized trajectories. Use MotionGenPlanConfig.time_dilation_factor<1.0 to slow down a
  planned trajectory. This is more robust than setting velocity_scale<1.0 and also allows for
  changing the speed of trajectories between planning calls
• curobo.util.logger adds logger_name as an input, enabling use of logging api with other
  packages.

Changes in default behavior

• Move CudaRobotModelState from curobo.cuda_robot_model.types to
  curobo.cuda_robot_model.cuda_robot_model
• Activation distance for bound cost in now a ratio instead of absolute value to account for very
  small range of joint limits when velocity_scale<0.1.
• TrajResult is renamed to TrajOptResult to be consistent with other solvers.
• Order of inputs to get_batch_interpolated_trajectory has changed.
• MpcSolverConfig.load_from_robot_config uses world_model instead of world_cfg to be
  consistent with other wrappers.

BugFixes & Misc.

• Fix bug in MotionGen.plan_batch_env where graph planner was being set to True. This also fixes
  isaac sim example batch_motion_gen_reacher.py.
• Add min_dt as a parameter to MotionGenConfig and TrajOptSolverConfig to improve readability
  and allow for having smaller interpolation_dt.
• Add epsilon to min_dt to make sure after time scaling, joint temporal values are not exactly
  at their limits.
• Remove 0.02 offset for max_joint_vel and max_joint_acc in TrajOptSolver
• Bound cost now scales the cost by 1/limit_range**2 when limit_range<1.0 to be robust to small
  joint limits.
• Added documentation for curobo.util.logger, curobo.wrap.reacher.motion_gen,
  curobo.wrap.reacher.mpc, and curobo.wrap.reacher.trajopt.
• When interpolation buffer is smaller than required, a new buffer is created with a warning
  instead of raising an exception.
• torch.cuda.synchronize() now only synchronizes specified cuda device with
  torch.cuda.synchronize(device=self.tensor_args.device)
• Added python example for MPC.