hghave: add some official category for known-bad and missing-good output

This will make it simple to tag output that are expected to changes. A simple
hghave integration seems enough but smarter behavior around these seems possible
in the future.

Differential Revision: https://phab.mercurial-scm.org/D9607

tests/hghave.py

@@ -14,6 +14,8 b" tempprefix = 'hg-hghave-'"
 checks = {
     "true": (lambda: True, "yak shaving"),
     "false": (lambda: False, "nail clipper"),
+    "known-bad-output": (lambda: True, "use for currently known bad output"),
+    "missing-correct-output": (lambda: False, "use for missing good output"),
 }
 
 try: