[ODE] New "slider2" joint

[Geoff Carlton]

I've been experimenting with joints, and I've come up with an 
alternative to the slider that seems to work better (at least, in the 
cases I'm interested in).  The reason I've been doing this is because 
there is a discrepancy between the suspension of a hinge2 vs slider - 
the hinge2 is much more stable when plugging in identical values.  By 
stable I mean that a box with 4 spheres using sliders tends to have a 
lot of "noise", or random bouncing, in addition to becoming a bit 
unstable at different cfm/erp values.

This "slider2" acts like a slider, but internally calls setBall2 and 
setFixedOrientation.  Thus, unlike the slider, it has 6 "nub"s.  It also 
has anchor points, like the hinge2, but no stops or motors (setBall2 
means that it tries to settle to a "stop" of 0.0 using the suspension 
cfm/erp).

Given the poor results I've been having with the normal slider, I'd 
recommend giving this alternative a try.  I'd also recommend having this 
joint  in the normal ode package, or at least in contrib.

Geoff