Improve assert patterns for bats tests so more contextual information is included

[Kami]

Right now we use pattern similar to the one below in some of our bats tests:

	SETUP_VENV_RESULTS=$(st2 run packs.setup_virtualenv packs=examples -j)
	run eval "echo '$SETUP_VENV_RESULTS' | jq -r '.result.result'"
	assert_success

        RESULTS=$(st2 rule list -p chatops -j)
        run eval "echo '$RESULTS' | jq -r '.[].ref'"
        assert_success

etc.

This pattern is far from ideal because if assertion fails (but the original command succeeds), it won't include any output of the original command (st2 run packs.setup_virtualenv packs=examples -j) which means it will be impossible to know why the tests have failed.

Only way to get more insight is by re-running those tests and that's very slow and painful process. In a lot of cases due to server not being available anymore, etc. this could mean wasting at least another 20 minutes of our valuable time.

We should establish the same pattern as we had with robot framework - always print command stdout, stderr and exit code.

Without that, debugging test failures is a nightmare and not an improvement over robot. In fact, I would even view it as a regression and a big step backward.

It could be as simply as adding bats utility function which runs a command, returns stdout and print stdout, stderr and exit code to console for debugging.

[Kami]

I propose something like this:

run_command_and_log_output() {
    output=$(exec $@)
    exit_code=$?

    # 1. Log the output
    >&2 echo "Stdout: ${output}"
    >&2 echo "Exit code: ${exit_code}"

    # 2. Return original command value
    echo ${output}
}

Then we would do something like:

RESULTS=$(run_command_and_log_output st2 rule list -p chatops -j)

instead of

RESULTS=$(st2 rule list -p chatops -j)

[Kami]

To give you an idea where the problem lies.

Here is current state of affairs.

Command / test fails, but nothing is logged due to the "anti-pattern" we use.

 ✗ packs.setup_virtualenv without python3 flags works and defaults to Python 2
   (in test file cli/test_pack_python3.bats, line 38)
     `SETUP_VENV_RESULTS=$(st2 run packs.setup_virtualenv packs=examples -j)' failed

1 test, 1 failure

With my proposed approach:

 ✗ packs.setup_virtualenv without python3 flags works and defaults to Python 2
   (in test file cli/test_pack_python3.bats, line 41)
     `SETUP_VENV_RESULTS=$(st2 run packs.setup_virtualenv packs=examples -j)' failed
     None
   
   Ran command: st2 run packs.setup_virtualenv packs=examples -j
   Stdout: 
   {
       "id": "5cdc58e0fb071a348b9cabce", 
       "parameters": {
           "packs": [
               "examples"
           ]
       }, 
       "result": {
           "exit_code": 1, 
           "result": "None", 
           "stderr": "st2.actions.python.SetupVirtualEnvironmentAction: DEBUG    Setting up virtualenv for pack \"examples\" (None)\nTraceback (most recent call last):\n  File \"/opt/stackstorm/st2/local/lib/python2.7/site-packages/python_runner/python_action_wrapper.py\", line 333, in <module>\n    obj.run()\n  File \"/opt/stackstorm/st2/local/lib/python2.7/site-packages/python_runner/python_action_wrapper.py\", line 192, in run\n    output = action.run(**self._parameters)\n  File \"/opt/stackstorm/packs/packs/actions/pack_mgmt/setup_virtualenv.py\", line 88, in run\n    no_download=no_download)\n  File \"/opt/stackstorm/st2/lib/python2.7/site-packages/st2common/util/virtualenvs.py\", line 86, in setup_pack_virtualenv\n    raise Exception(msg)\nException: Pack \"examples\" is not installed. Looked in: /opt/stackstorm/packs\n", 
           "stdout": ""
       }, 
       "status": "failed"
   }
   Exit code: 1

This would have probably saved me at least 4 hours today.