[ODE] Improved QuickStep stability...?

[Russ Smith]

> Don't hit me -- I added random constraint reordering to QuickStep
> and it helped stability a lot

wow, this does help a lot. i've checked it in and made it the default
for now. there are #defines there now to enable the three 'improvement
methods': croteam's reordering, warm starting, and random reordering.

i've also updated test_crash to be able to use stepfast or quickstep,
and selected quickstep by default.

the source release has been rebuilt.

> warm-starting seems to play havoc with test_crash's buggy stability.

yes: the problem seems to be with the friction approximation #1 and the
very large mu (friction) value used for the wheel-ground interaction in
that demo. the problem is somewhat similar to the problem i saw with
motors, i think. warm starting the contacts too should help, eventually.

russ.

-- 
Russell Smith
http://www.q12.org