What's a reasonable size ratio for toom32?

[Paul Zimmermann]

       Hi Niels,

> Not sure what would be the appropriate way to benchmark; could look at a
> range of unbalancedness ratio, at one or a few fixed sizes relevant to
> toom22 and toom32, or look at fixed ratio, perhaps 1/2, over a range of
> sizes.

I suggest taking fixed non-rational ratios, for example sqrt(2), sqrt(3),
sqrt(5), and comparing the old and new code over the range of sizes where
this code is used.

Paul