fix: defer triggered actions until after activation frame

[thomassedlmayer]

This changes storyboard event handling so triggered actions are queued instead of being applied in the same frame that activates their start condition. The activation frame now remains unchanged, and action effects become visible in a following simulation step.

The current "immediate application" concept surfaced an exceeded MAE tolerance only for the falling-edge speed trigger validation case. I think this issue didn't appear for other similar test cases because the extra 1-frame lag between reference and tool trajectory accidentally didn't exceed the error tolerances while it did for the falling-edge case.

Scheduling options:

• Immediate application
  • Trigger fires at t.
  • Trajectory is attached at t.
  • First vertex can be visible at t.
• Apply pending actions after _advance()
  • Trigger fires at t.
  • Frame t stays unchanged.
  • Trajectory is attached at t + dt.
  • First vertex becomes visible around t + 2dt.
• Apply pending actions before _advance()
  • Trigger fires at t.
  • Frame t stays unchanged.
  • Trajectory is attached before advancing to t + dt.
  • Trajectory motion is visible at t + dt, but leading vertices may be skipped (depending on frame rate).

It would also be reasonable to increase the tolerated alignment lag from 1 to 2 in the corresponding test cases. With that, any scenario engine using "immediate application" would be tolerated by the validation suite.