Switching default integrator for multiple variable to quadcc

[Rik]

4/18/11

All,

The dblquad and triplequad functions for double and triple integrals are
slow because the algorithm involves repeatedly calling a 1-dimensional
integrator and adding up small slices.  The default 1-D integrator
previously was quadgk, but my testing shows that quadcc is *much* faster.
Accordingly, I have switched the default backend integrator to quadcc on
the development branch.

Test results:

For dblquad (@(x,y) exp(-x.^2 - y.^2) , -1, 1, -1, 1, 1e-6) executed 30
times on uniprocessor @ 1200 MHz.  All times in seconds.

quadcc: 0.1918
quadl:  4.262
quadgk: 8.868
quadv:  19.65

Cheers,
Rik