Boundary correction

[kincaidkc]

Closes #5, where the laser power is increased as the center crosses the boundary to maintain the total setpoint power.

This works in several steps, outlined as follows:

1. The heat source, qDot, is given a finite value only within a set distance (2 sigma) of the beam center, and is set to zero elsewhere. A D4sigma helper function is added to each heatSourceModel derived class to assist in this calculation.
2. The algorithm tests if the total resolved power is greater than some small value. If it is not, this means either that the beam is off, or the center is greater than 2 sigma to the outside of the boundary, thus qDot has a zero value in every cell. If the resolved power is less than the arbitrarily small value, no normalization occurs. If it is greater, the calculation proceeds.
3. The distance of the beam center from the boundary is then calculated by comparing the beam position with the locations of the points forming the outline of the patch on which the scan occurs. If a boundary point less than 2 sigma from the beam position is located, the loop breaks and no normalization occurs. Otherwise, qDot is corrected such that the total power is equal to the setpoint power.

One new mandatory dictionary entry, scanPatchName, is added to heatSourceDict. An additional optional control, normalize, is added to allow users to disable the normalization procedure completely in cases where a high-resolution mesh is used.

[streeve]

Note - "closes #" or "fixes #" will automatically link and close issues

[colemanjs]

Alternative adaptive sampling method fixed in #15