Ignore errors based on stdout/stderr

Currently, we have a few options to decide when an operation is called success:

1. The exit_code can be chosen to whatever integer and that would be called success
2. We can just always say something is successful
3. And other features such as postconditions and timeouts

What I'm interested in is distinguishing whether something is successful based on the stdout/stderr of the operation. For example, if you run the following command twice:

iptables --new foo

you'll get the following error message the second time in stderr:

iptables: Chain already exists.

with an exit code of 1. The problem is that a lot programs don't really have nuance with their exit codes. It's often 0 for success, and then 1 or 2 or whatever for failure.

So what would be useful is to have a decision of pass/failure based on some string matching of the stdout/stderr.

I think ansible might have this feature already, if that helps your decision.

It could be as simple as pass_on_match=<regexp pattern>.

Start by reviewing the existing exit_code, always-success, postcondition, and timeout options described in the issue, along with how operation stdout and stderr are exposed. Done means the project has a decided behavior for matching output, including the proposed pass_on_match regular-expression form, and distinguishes matching successes from failures.