Trajectory Analysis improvements

[sixpearls]

Includes #41, or can target that branch.

So far:

• adds support for specifying t0
• adds support for non-scalar trajectory_output
• flag for storing separate event updates
• flag for continuing/resetting step size after event (SciPy)
• support for resampling a trajectory

Potentially want to use this PR to extend derivative calculations

• making adjoint computational dynamically generated
• support for forward sensitivity (also dynamically generated)
• higher order derivatives by stacking (e.g., forward-over-reverse)
• improve result interpolant api, convenience function for re-sampling, etc.

[sixpearls]

Future work from original description tracked in #53, this MR is ready to go.