Implement reduceTransitions to allow efficient accumulation over transitions

[merkste]

The folding operation reduce is a principle iteration method for sequences of any kind. It takes an initial value and enables the accumulation of the results of each step. This PR adds reduceTransitions(state, init, ObjTransitionFunction) to enable iteration over the transitions of some model and accumulating the results, e.g., consider the following code change:

public default double sumOverTransitions(final int state, final TransitionToDoubleFunction f)
{
    class Sum {
        double sum = 0.0;
        void accept(int s, int t, double d)
        {
            sum += f.apply(s, t, d);
        }
    }
    Sum sum = new Sum();
    forEachTransition(s, sum::accept);
    return sum.sum;
}

to

public default double sumOverTransitions(final int state, final TransitionToDoubleFunction f)
{
    return reduceTransitions(state, 0.0, (r, s, t, d) -> r + f.apply(s, t, d));
}

In each step, the function f is applied to the current transition. The result of each step is the sum of all intermediate values up to the current transition.

The implementation of reduceTransitions should not rely on forEachTransition, as the access to a member-variable of some function (see the example above) is generally slower than to a temporary variable in reduceTransitions. Also, as functions passed to reduce are often clean, i.e., they do not reference external objects/variables, they could be efficiently inlined by the virtual machine (JVM does not yet apply this optimization).

We could go even futher by considering forEachTransition a special case of reduceTransitions that simply dismisses the intermediate results. This should yield equal performance in theory, but would require some tests to confirm that no performance hit occurs in practice.

[merkste]

I completed the implementation for DTMCs and CTMCs and added support for MDPs. The last commit implements forEachTransitionDo in terms of reduce.

[merkste]

@davexparker , @kleinj : Now with 4.5 out, what do you think?