Make the LCP internal error printout less aggressive

[mxgrey]

A line in the LCP solver may print out excessive warnings over minor imprecision. This causes unnecessary terminal spam that looks like this:

ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)

ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)

ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)

ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)

ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)

ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)

ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)

ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)

ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)

ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)

ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)

ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)

ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)

ODE Message 3: LCP internal error, s <= 0 (s=0.0000e+00)

ODE Message 3: LCP internal error, s <= 0 (s=-1.5440e-13)

ODE Message 3: LCP internal error, s <= 0 (s=-6.4683e-13)

These messages often print even when nothing is really wrong with the simulation. I've wrapped the printout of this message in an if-statement that should make the LCP solver less aggressive when it comes to printing this out.

---

Before merging a pull request

• Set version target by selecting a milestone on the right side
• Summarize this change in CHANGELOG.md

[jslee02]

Theoretically, the "s" value should always be positive. It's undefined behavior otherwise. However, it seems the tiny values are negligible.

For the case of s <= 0, we may want to fall back to using an iterative constraint solver, which is less accurate but more robust, once it's available. But let's stick to Dantzig until then.

[mxgrey]

It wouldn't surprise me if these tiny "failures" boil down to floating point precision issues.

[scpeters]

Another approach is calling dSetMessageHandler to completely silence the ODE messages:

• https://bitbucket.org/osrf/gazebo/pull-requests/2512/create-ode-parameter-to-silence-solver/diff