Develop the learner of NPG algorithm

[wenzhangliu]

• Initialize the NPG learner. Add the optimizer, scheduler, loss, gamma, etc.
• The update method.

[wenzhangliu]

The agent and learner modules of the NPG algorithm are successfully added. However, before running the NPG example, the configuration files (xuance/configs/npg/xxx.yaml) must be added. @XiangDuojie

import xuance

runner = xuance.get_runner(method='npg',
                           env='classic_control',
                           env_id='CartPole-v1',
                           is_test=False)
runner.run()