Seven Over Four

Sutton and Barto's 4x4 gridworld. The two shaded corners are one terminal state, reward is $-1$ on every transition, $\gamma = 1$, and moves that would leave the grid keep the agent where it is. The policy is equiprobable random, so each of up, down, left, right has probability 0.25.

After the first sweep, $v_1(s) = -1$ for all fourteen nonterminal states and $v_1(\text{T}) = 0$.

Compute $v_2(1)$.