Instead of monkeypatching I just reduced global gap limits. The slowest test is down to 3 from 8 seconds (and maybe it can be improved more), several other tests are faster too.
Interestingly, tests get even faster if we don't restore the global gap limits, and they still pass.
(I changed the rather obscure og prefix into orig.)
This fixes #157 (or at least improves it).
Instead of monkeypatching I just reduced global gap limits. The slowest test is down to 3 from 8 seconds (and maybe it can be improved more), several other tests are faster too.
Interestingly, tests get even faster if we don't restore the global gap limits, and they still pass.
(I changed the rather obscure
og
prefix intoorig
.)