states, actions, rewards = generate_episodes(nb, height, width, T)
seq = episodes2seq(states, actions, rewards, lookahead_delta=T)
s, a, r, lr = seq2episodes(seq, height, width, lookahead=True)
states, actions, rewards = generate_episodes(nb, height, width, T)
seq = episodes2seq(states, actions, rewards, lookahead_delta=T)
s, a, r, lr = seq2episodes(seq, height, width, lookahead=True)