JUnit reporter uses only 3 decimal places when reporting durations

We used to use whatever precision we ended up having from C++'s
stdlib. However, some relatively popular tools, like Jenkins,
use Maven SureFire XML schema to validate JUnit test reports, and
Maven SureFire schema requires the duration to have at most 3
decimal places.

For compatibility, the JUnit reporter will now respect this
limitation.

Closes #2221

scripts/approvalTests.py

@@ -28,7 +28,10 @@ filelocParser = re.compile(r'''
 ''', re.VERBOSE)
 lineNumberParser = re.compile(r' line="[0-9]*"')
 hexParser = re.compile(r'\b(0[xX][0-9a-fA-F]+)\b')
-durationsParser = re.compile(r' time="[0-9]*\.[0-9]*"')
+# Note: junit must serialize time with 3 (or or less) decimal places
+#       before generalizing this parser, make sure that this is checked
+#       in other places too.
+junitDurationsParser = re.compile(r' time="[0-9]*\.[0-9]{3}"')
 sonarqubeDurationParser = re.compile(r' duration="[0-9]+"')
 timestampsParser = re.compile(r'\d{4}-\d{2}-\d{2}T\d{2}\:\d{2}\:\d{2}Z')
 versionParser = re.compile(r'Catch v[0-9]+\.[0-9]+\.[0-9]+(-develop\.[0-9]+)?')
@@ -138,7 +141,7 @@ def filterLine(line, isCompact):
     line = hexParser.sub("0x<hex digits>", line)
 
     # strip durations and timestamps
-    line = durationsParser.sub(' time="{duration}"', line)
+    line = junitDurationsParser.sub(' time="{duration}"', line)
     line = sonarqubeDurationParser.sub(' duration="{duration}"', line)
     line = timestampsParser.sub('{iso8601-timestamp}', line)
     line = specialCaseParser.sub('file:\g<1>', line)