Add optional argument forwarding to predict/update using variadic templates

[mherb]

Example:

// Time-step
float dt = 0.001;
// Pass time-step as optional argument, which is forwarded to sys.f(x,u,dt)
filter.predict(sys, u, dt);

However, this would require to remove virtual-ness of fand h in System/Measurement-Models to allow custom arguments in each model.

[wujilingfeng]

Example:

// Time-step
float dt = 0.001;
// Pass time-step as optional argument, which is forwarded to sys.f(x,u,dt)
filter.predict(sys, u, dt);

However, this would require to remove virtual-ness of fand h in System/Measurement-Models to allow custom arguments in each model.

Why did UnscentedKalmanFilter not add this parameter?