Effort limits for GcsTrajectoryOptimization

[aelvio]

Similar to #22500, it would be useful to have a function where you could impose a MultibodyPlant's effort limits onto individual subgraphs or regions of a GcsTrajectoryOptimization problem. An interface like subgraph.AddEffortBoundsAtNormalizedTimes(plant, s, plant_context) would work great for us, where the normalized time is within each segment. I believe the implementation would make use of #21981 but wouldn't know how to say more about implementation details.

#22500 has already been very useful to us for coming up with optimally-fast trajectories that don't violate effort limits, but we'd like to switch from KinematicTrajectoryOptimization to GcsTrajectoryOptimization in order to extend that work to scenarios with multiple options. @RussTedrake asked me to open an issue here.

In particular I would love to be able to impose different MultibodyPlant's on different regions/subgraphs of configuration space - I have a scheme that you could get a "poor man's planning through contact" for certain problems (including ours) by doing this.