perf: run the gc before each run

The python garbage collector is a large source of performance troubles, we run
it right before the timed section to reduce the change for the gc to add noise
to the benchmark.

contrib/perf.py

@@ -581,6 +581,7 b' def _timer('
         if setup is not None:
             setup()
         with context():
+            gc.collect()
             with prof:
                 with timeone() as item:
                     r = func()