provide a sample test to benchmark the impact of caching the dates

As a reference on a laptop with :
Intel(R) Core(TM) i7-2630QM CPU @ 2.00GHz
Python 2.7.7rc1
Debian SID - Linux 3.14-1-amd64 matomo-org#1 SMP Debian 3.14.4-1 (2014-05-13) x86_64 GNU/Linux

The results are :
cache   : 647.914 ms
nocache : 21440.542 ms

tests/debug/bench_date_cache.py

@@ -0,0 +1,46 @@
+#!/usr/bin/env python
+
+import datetime
+import time
+from collections import OrderedDict
+
+date_format = '%d/%b/%Y:%H:%M:%S'
+
+def timing(f):
+    def wrap(*args):
+        time1 = time.time()
+        ret = f(*args)
+        time2 = time.time()
+        print '%s\t: %0.3f ms' % (f.func_name, (time2-time1)*1000.0)
+        return ret
+    return wrap
+
+@timing
+def nocache():
+    file = open("timestamps.log", 'r')
+    for lineno, line in enumerate(file):
+        date_string, timezone_string = line.split()
+
+        date = datetime.datetime.strptime(date_string, date_format)
+        timezone = float(timezone_string)
+
+        date -= datetime.timedelta(hours=timezone/100)
+
+@timing
+def cache():
+    cache_dates = OrderedDict()
+    file = open("timestamps.log", 'r')
+    for lineno, line in enumerate(file):
+        date_string, timezone_string = line.split()
+        key = line
+
+        date = cache_dates.get(key)
+        if not date:
+            date = datetime.datetime.strptime(date_string, date_format)
+            timezone = float(timezone_string)
+
+            date -= datetime.timedelta(hours=timezone/100)
+            cache_dates[key] = date
+
+cache()
+nocache()