Cuda kernel init errors caused by massive number of objects being spawned at 0 during reconfiguration

[StoneT2000]

Issue found by @arth-shukla . Currently by default if no initial pose is provided all objects built will be first spawn at 0 and then a single physx step is taken to initialize the simulation. If there are too many objects this causes too many collisions initially which can break the simulation / lead to illegal memory accesses.

One potential solution is to by default during reconfiguration set the pose of all objects to different Z-heights, which usually will not conflict with other objects in most user-created tasks.