Implementing Policy gradient in seqGAN

Hi,

I want to implement the MC search with roll-out policy in the SeqGAN code here (as described in the [original paper](https://arxiv.org/abs/1609.05473)). My idea was to create a custom Helper that would help implement the roll-out policy. However I found the [SeqPGAgent](https://texar.readthedocs.io/en/latest/code/agents.html#texar.tf.agents.SeqPGAgent) code which seems to be usable here. Can you please advice on whether I can use the SeqPGAgent here to implement this rollout policy? Thanks!

Start with the SeqPGAgent documentation entry point and compare its documented behavior with the SeqGAN rollout policy described in the linked original paper. Then inspect the repository's SeqGAN implementation and determine whether a custom Helper or SeqPGAgent is the appropriate integration point; done means the project has a decided implementation path, not just usage advice.

Written by the indexing model from the issue text.