[MLA-1809] catch mismatched observation sizes

[chriselion]

Proposed change(s)

Even though we check and assert on the Unity side, it's possible to get observations of different size send over. This will cause a numpy exception when trying to pack them into an array:

2021-03-01 20:07:58 ERROR [subprocess_env_manager.py:226] UnityEnvironment worker 0: environment raised an unexpected exception.
TypeError: float() argument must be a string or a number, not 'google.protobuf.pyext._message.RepeatedScalarContainer'

The above exception was the direct cause of the following exception:

Traceback (most recent call last):
  File "/Users/chris.elion/code/ml-agents/ml-agents/mlagents/trainers/subprocess_env_manager.py", line 208, in worker
    env.reset()
  File "/Users/chris.elion/code/ml-agents/ml-agents-envs/mlagents_envs/environment.py", line 309, in reset
    self._update_state(rl_output)
  File "/Users/chris.elion/code/ml-agents/ml-agents-envs/mlagents_envs/environment.py", line 290, in _update_state
    self._env_state[brain_name] = steps_from_proto(
  File "/Users/chris.elion/code/ml-agents/ml-agents-envs/mlagents_envs/timers.py", line 305, in wrapped
    return func(*args, **kwargs)
  File "/Users/chris.elion/code/ml-agents/ml-agents-envs/mlagents_envs/rpc_utils.py", line 304, in steps_from_proto
    _process_rank_one_or_two_observation(
  File "/Users/chris.elion/code/ml-agents/ml-agents-envs/mlagents_envs/timers.py", line 305, in wrapped
    return func(*args, **kwargs)
  File "/Users/chris.elion/code/ml-agents/ml-agents-envs/mlagents_envs/rpc_utils.py", line 265, in _process_rank_one_or_two_observation
    np_obs = np.array(
ValueError: setting an array element with a sequence.

This change adds an extra check that the observations all have the shape of the observation spec:

2021-03-03 17:42:32 ERROR [subprocess_env_manager.py:226] UnityEnvironment worker 0: environment raised an unexpected exception.
Traceback (most recent call last):
  File "/Users/chris.elion/code/ml-agents/ml-agents/mlagents/trainers/subprocess_env_manager.py", line 208, in worker
    env.reset()
  File "/Users/chris.elion/code/ml-agents/ml-agents-envs/mlagents_envs/environment.py", line 309, in reset
    self._update_state(rl_output)
  File "/Users/chris.elion/code/ml-agents/ml-agents-envs/mlagents_envs/environment.py", line 290, in _update_state
    self._env_state[brain_name] = steps_from_proto(
  File "/Users/chris.elion/code/ml-agents/ml-agents-envs/mlagents_envs/timers.py", line 305, in wrapped
    return func(*args, **kwargs)
  File "/Users/chris.elion/code/ml-agents/ml-agents-envs/mlagents_envs/rpc_utils.py", line 295, in steps_from_proto
    raise ValueError(
ValueError: Observation at index=0 for agent with id=7 didn't match the ObservationSpec. Expeceted shape (8,) but got (9,).

Useful links (Github issues, JIRA tickets, ML-Agents forum threads etc.)

https://forum.unity.com/threads/walker-demo-adding-blockcontact-observation-valueerror-setting-an-array-element-with-a-sequence.1065587/#post-6895055

Types of change(s)

• Bug fix

Checklist

• Added tests that prove my fix is effective or that my feature works
• Updated the changelog (if applicable)

[vincentpierre]

I think this is better now.