As the plugin processes test results from the phases, it merges the results together. If there are passed scenarios and failed, the overall job is marked as unstable, if all jobs fail then it is marked as failure.

The inconsistency is:

a) if the first cucumber test to finish fails and subsequent tests pass, the overall pipeline turns red.

vs

b) if the first cucumber test passes and subsequent ones fail, the pipeline turns yellow.

It is desirable that the result of the pipeline be consistent, regardless of the order the tests finish in.

Alternatively, if we could pass in a parameter to avoid any manipulation of the overall build result from within this plugin, we can handle manipulating the build result as part of a separate phase after the tests complete.