Use Pose instead of sapien.Pose in update_camera_configs_from_dict()

[SoshinK]

If I want to change camera's pose and have JSON serializable args, I have to pass a new pose as a list, according to

ManiSkill/mani_skill/sensors/camera.py

Lines 107 to 109 in dbce310

	# for json serailizable gym.make args, user has to pass a list, not a Pose object.
	if "pose" in v and isinstance(v["pose"], list):
	v["pose"] = sapien.Pose(v["pose"][:3], v["pose"][3:])

Not very clear, how can I set different camera poses in different parallel environments. Isn't it more convenient to pass a list of lists with camera poses and cast it to Pose, i.e. use v["pose"] = Pose.create_from_pq(v["pose"][:, :3], v["pose"][:, 3:]) instead of v["pose"] = sapien.Pose(v["pose"][:3], v["pose"][3:])?