Added support for minibatch in PPO process_fn (#1168)

Closes #1164

In PPOPolicy, the method `process_fn()` now computes `logp_old` in
minibatch instead of all at once.

---------

Co-authored-by: Michael Panchenko <[email protected]>

tianshou/policy/modelfree/ppo.py

@@ -137,8 +137,11 @@ def process_fn(
             self._buffer, self._indices = buffer, indices
         batch = self._compute_returns(batch, buffer, indices)
         batch.act = to_torch_as(batch.act, batch.v_s)
+        logp_old = []
         with torch.no_grad():
-            batch.logp_old = self(batch).dist.log_prob(batch.act)
+            for minibatch in batch.split(self.max_batchsize, shuffle=False, merge_last=True):
+                logp_old.append(self(minibatch).dist.log_prob(minibatch.act))
+            batch.logp_old = torch.cat(logp_old, dim=0).flatten()
         batch: LogpOldProtocol
         return batch