LearnableValueApproximationIfc

A value function that improves from observed experience.

The seam exists so that a LookaheadPolicy holding one of these becomes a learning rule by forwarding ManagedPolicyIfc.onTransition to update — one class rather than a new concept.

The epoch loop delivers the transitions and the rewards it needs: M1 steps 7b and 7c wired §4.10.2's steps 2 and 4, and §8.2.9 measures the hooks firing. No shipped class forwards them to update — that adapter is the one piece still unwritten, and writing it is not the same problem as choosing a fitting algorithm, which is out of scope (§1.2).