I need to be able to specify which warnings to include based on a pattern,
similar to exclude pattern from issue 2359. In my case it is much simpler to
specify what to include into report than specifying what to exclude (almost
I can implement it myself, just want to make sure it gets upstream.