Follow up on #3307

The NoteSuppressedWarnings detector which is a NonReportingDetector reports now warnings.
This is inconsistent, and as every inconsistency this has side effects.

In Eclipse we used BugInstance.getDetectorFactory() to retrieve information about the spotbugs plugin that contributes this detector, and if not available (which is the case for NonReportingDetector), we were trying to scan the DetectorFactory for the matching bug patterns. However, we've restricted that to only factories which match to DetectorFactory.isReportingDetector()condition. Again, this is not the case forNonReportingDetector`:

We can add a workaround in Eclipse plugin like below, but it would be better if either the bug instances would have the factory added or the NoteSuppressedWarnings would be a reporting detector. Later seem to be a problem as the comment says it has to be executed before all "reporting" detectors and so simply removing implements NonReportingDetector from it causes a cycle & analysis crash.

diff --git a/eclipsePlugin/src/de/tobject/findbugs/reporter/MarkerReporter.java b/eclipsePlugin/src/de/tobject/findbugs/reporter/MarkerReporter.java
index 70259a2..e006d6c 100644
--- a/eclipsePlugin/src/de/tobject/findbugs/reporter/MarkerReporter.java
+++ b/eclipsePlugin/src/de/tobject/findbugs/reporter/MarkerReporter.java
@@ -59,2 +59,3 @@
 import edu.umd.cs.findbugs.config.UserPreferences;
+import edu.umd.cs.findbugs.detect.NoteSuppressedWarnings;
 
@@ -215,3 +216,6 @@
                 if (!df2.isReportingDetector()) {
-                    continue;
+                    // NoteSuppressedWarnings reports bugs even if it says it doesn't ...
+                    if (!NoteSuppressedWarnings.class.getName().equals(df2.getFullName())) {
+                        continue;
+                    }
                 }

@gtoison : would be nice if you could look at it.