Fix CI errors due to pyglet, zipp, mock, and gym

[muupan]

• Use newer gym because old gym does not work with pyglet>=1.4.
• Modify some example scripts that are not compatible with newer gym.
• Use zipp==1.0.0 in flexCI because new zipp does not work with Python 3.5.
• Remove mock from dependency as we do not support py2 now.

[muupan]

/test

[pfn-ci-bot]

Successfully created a job for commit 890f858:

• Dashboard for commit 890f858

[muupan]

/test

[pfn-ci-bot]

Successfully created a job for commit 65534f0:

• Dashboard for commit 65534f0

[toslunar]

LGTM

[toslunar]

https://travis-ci.org/chainer/chainerrl/jobs/649363317#L2503

$ ./test_examples.sh -1
Running example tests: examples_tests/atlas/test_soft_actor_critic.sh
Output files are saved in /tmp/tmp.cdt7eztf9w/atlas/soft_actor_critic/20200212T100236.429911
/home/travis/virtualenv/python3.6.7/lib/python3.6/site-packages/gym/logger.py:30: UserWarning: WARN: Box bound precision lowered by casting to float32
  warnings.warn(colorize('%s: %s'%('WARN', msg % args), 'yellow'))
Traceback (most recent call last):
  File "examples/atlas/train_soft_actor_critic_atlas.py", line 243, in <module>
    main()
  File "examples/atlas/train_soft_actor_critic_atlas.py", line 123, in main
    timestep_limit = sample_env.spec.tags.get(
AttributeError: 'EnvSpec' object has no attribute 'tags'
The command "./test_examples.sh -1" exited with 1.

due to the version of gym?

[muupan]

/test

[pfn-ci-bot]

Successfully created a job for commit 73a1a44:

• Dashboard for commit 73a1a44

[muupan]

/test

[pfn-ci-bot]

Successfully created a job for commit 7fb7212:

• Dashboard for commit 7fb7212

[muupan]

/test

[pfn-ci-bot]

Successfully created a job for commit 54b49c9:

• Dashboard for commit 54b49c9

[muupan]

@toslunar I added changes to fix errors caused by gym==0.16.0. Now CI is all ok. Can you check the additional changes and merge if it's ok?

[toslunar]

I confirmed env.spec.max_episode_steps is available with gym 0.9.7, the minimum requirement by ChainerRL.

[toslunar]

LGTM