Created originally on Bitbucket by _stian (Stian Andreassen)
Was already merged in Bitbucket before import, marked as merged by the import user
Improves just about all rbigint operations. Upto twice as fast (power of two even better)
Tests passes.
(Update with an important bugfix for SUPPORT_INT128 = False)
(Update 2, fix those things arigo pointed out, it seems to build now using a 32bit chroot of archlinux)
(Update 3, slightly more effective _normalize, pow, revert tc_mul to the more effective version if anyone wants to play with it, and remove a modification of rffi.TYPES from rbigint.py)