Adding Failed exception to manage maxfail behavior

[jespino]

As I explained in the issue #2832, I'm not sure that the behavior is incorrect by itself, but have sense both options, interrupt return, and failed return. Anyway, I prefer to have failed return and leave "interrupt" to explicit user interruptions, like Ctrl-C.

[RonnyPfannschmidt]

good initial work 👍

due to the nature of the bugfix here i believe its better to target the features branch here

please also add a test that demonstrates the behaviour in the testsuite as well as a changelog entry

[jespino]

Added a changelog entry, changed the branch to merge to feature, and the current tests are changed to validate the new behavior. The existing test already test the behavior.

[coveralls]

Coverage decreased (-0.01%) to 92.658% when pulling 00d3abe on jespino:fix/2832 into 71c76d9 on pytest-dev:features.

[coveralls]

Coverage decreased (-0.01%) to 92.658% when pulling f690161 on jespino:fix/2832 into 71c76d9 on pytest-dev:features.

[nicoddemus]

The existing test already test the behavior.

I agree that's enough, unless @RonnyPfannschmidt has something else to test in mind.

The two failing tests are unrelated (#2843).

[coveralls]

Coverage decreased (-0.01%) to 92.658% when pulling e81b275 on jespino:fix/2832 into 71c76d9 on pytest-dev:features.