Fix the planar hand and pushing examples

[aykut-tri]

This PR proposes two small fixes for the following issues:

• The BatchQuasistaticSimulator::CalcBundledABcTrj method assumes that the last time step is removed from the state trajectory. This is respected by irs_mpc2 but was not by rs_mpc, causing the planar hand examples to fail
• The time step size was not set for the IRS MPC parameters in the planar pushing example, and it being NaN by default was failing the QP that is solved to take simulation steps

---

This change is 

[aykut-tri]

Sure thing. I also guessed irs_mpc2 would make irs_mpc obsolete at some point but it was a super easy fix, so I wanted to push it regardless.

The run_planar_pushing2 example was failing because h was not set in the irs_mpc2 parameters, and it is NaN by default. Also, publish_trajectory at the end was not working since h is not passed. So, I just fixed those two issues in the example script.

[aykut-tri]

Gotcha, I'll let you guys decide on whether to merge this PR. I just wanted to record the issue/solution in case our future selves or other current users want to run the planar examples out of the box.

[pangtao22]

closed by #61