benchmarks: fix regexp for parsing code size results

Accept a '.' in the benchmark name which is used for .o and .dylib files

Author: eeckstein

benchmark/scripts/compare_perf_tests.py

@@ -308,7 +308,7 @@ def _reset(self):
 
     # Parse lines like this
     # #,TEST,SAMPLES,MIN(μs),MAX(μs),MEAN(μs),SD(μs),MEDIAN(μs)
-    results_re = re.compile(r'([ ]*\d+[, \t]*\w+[, \t]*' +
+    results_re = re.compile(r'( *\d+[, \t]*[\w.]+[, \t]*' +
                             r'[, \t]*'.join([r'[\d.]+'] * 6) +
                             r'[, \t]*[\d.]*)')  # optional MAX_RSS(B)