[PATCH 3 of 3] tests: make a stab at approximating wall-clock times

[Bryan O'Sullivan]

# HG changeset patch
# User Bryan O'Sullivan <bos at serpentine.com>
# Date 1451973286 28800
#      Mon Jan 04 21:54:46 2016 -0800
# Node ID 0486b6c4d3568ecb7d4e3f8190d200455fb53650
# Parent  bc79c418fbf83ba8016ff6838ddd2649c144bd5c
tests: make a stab at approximating wall-clock times

Vaguely empirical observations:

* ".py" tests are about an order of magnitude faster than ".t" tests

* dividing size by 1000 gives an approximation to wall-clock
  run time (in seconds) that is not completely ridiculous.

diff --git a/tests/run-tests.py b/tests/run-tests.py
--- a/tests/run-tests.py
+++ b/tests/run-tests.py
@@ -1861,7 +1861,9 @@ class TestRunner(object):
                     for kw, mul in slow.items():
                         if kw in f:
                             val *= mul
-                    perf[f] = val
+                    if f.endswith('.py'):
+                        val /= 10.0
+                    perf[f] = val / 1000.0
                     return perf[f]
             tests.sort(key=sortkey)