Add task benchmarking utilities

[knicked]

This PR adds a new package to the ros2 workspace responsible for providing a benchmark utility for performing tasks in the manipulation demo scene. Currently features two very simple scenarios, with an interface allowing for easily adding new scenarios.

Usage

First build and source the ros2_ws workspace, then launch the simulation as in here

Then run either

ros2 run rai_mani benchmark

which will run an example benchmark on a RAI conversational agent and show its results in the terminal, or

ros2 run rai_mani showcase

which will run the same benchmark, but with the manipulator doing scripted movements with no agent involved.