Make the evaluators inside a recurrent goup true evaluator, meaning that their evaluation results are incorporated into the whole evaluator result.

1. v2.layer.parse_network does not correctly handle the generation output.
2. GatherAgentLayer does not correctly handle generation output when batch_size > 1
3. Fix CustomStackTrace for rnn group

For example:
layer.printer(input=[x, y], format="x=%s y=%s")

No longer need to specify target_inlinks or is_seq.

No longer need to use SubsequenceInput. The framework will detect.

It seems that is not used.