The groovy plugin.xml extractor parses out .java and .groovy mojos, which causes duplicate conflicts

The groovy plugin.xml descriptor extractor parses java and groovy sources, so that groovy mojos can properly pick up javadoc annotations from java base classes. But it also picks up mojos implemented in java too, which causes the maven-plugin-plugin to detect a 2 mojos (one extracted from the java extractor, and another from the groovy extractor).

The groovy extractor should not include mojo descriptors for java sources, only groovy sources... yet it should still process both groovy and java sources for base-class annotation inclusion.