Add the ability to control the random seed used by regrtest.py -r. This
patch adds a --randseed option, and makes regrtest.py -r indicate what
random seed it's using so that that value can later be fed back to
--randseed. This option is useful for tracking down test order-related
issues found by make buildbottest, for example.

+1 on the idea.

I'm not sure I understand the patch, though.  If the line:

random_seed = int(1000000 * random.random())

produces a random_seed of 0 (or if randseed=0 is supplied as an option), 
it looks as though random.seed is never called;  is this intentional?

Also, not that it really matters, but why do you use the expression above 
instead of the simpler random.randrange(1000000)?

The "if rand_seed:" bit was a relic from a previous iteration; fixed.
The only reason I didn't use randrange() is that I didn't see it; fixed.

+1 here too.

Updated patch looks good to me.

Committed as r70672 (trunk) and r70673 (py3k). Thanks for the quick review.

Should this get a Misc/NEWS entry?