Include integrate functionality in ForestClaw

[donnaaboise]

This added @cburstedde integrate functionality to ForestClaw.

• See simple example in swirl. This example integrates a ray that starts at the origin (lower left corner of swirl domain) and ends along quarter circle in x>0, y>0. Radius of circle is set to .25, so integral along ray (which now only computes length of ray) is easy to compute.
• Code checks intersection of simple ray (arrow) with p4est quad, and determines if (1) ray misses quad entirely, (2) quad enters and exits quad, (3) ray enters quad but ends inside quad, (4) ray starts inside quad and exits.
  The length of the ray is computed by compute length of any segment of the intersection of ray with quad.

Several outstanding issues :

• What happens if a general ray (possible parameterized curve) does not intersection coarse level quad. Will finer levels quad in this coarser quad be tested?
• Direction of integration should be taken into account
• Interesting to see how to compute intersection with generalized parameterized curves.

[cburstedde]

This PR does not build on top of our latest integrate code. I think it will fit better if we were to allow to create a PR for our ray integration code first, merge it into develop, and then consider these commits?

[cburstedde]

You will find a lot more recent code on my bitbucket repository. We'll make a PR here once Hannes has included the intersection test that I had introduced on slack.