Function predict_kinematic_state 

pub fn predict_kinematic_state(
    state: &mut KinematicState,
    covariance_m2: &mut Vec<Vec<f64>>,
    dt_s: f64,
    active_ambiguity_ids: &[String],
    config: &KinematicConfig,
) -> Result<(), KinematicSolveError>

Predict one kinematic PPP EKF state and covariance forward by dt_s.

The function updates the state in place, resizes ambiguity states to the sorted active_ambiguity_ids, assigns configured variance to new ambiguities, drops inactive ambiguities, applies the configured motion model, and inflates the covariance by process noise scaled by elapsed time.