Polish up came config

Close #13

Until now, given real upper bounds (bx, by), with underlying
canvas size (sx, sy), coordinates bx and by would map directly
to sx and sy, thus out of the underlying discrete canvas.

This created some problems, for example a shape sized 1.0 on
one of the dimensions would not fit a RealCanvas of size 1.0 on
the same dimension.

With this change, coordinates equal to the upper bounds will
map to the last pixel of the canvas, thus bx -> (sx - 1) and
by -> (sy - 1).

One needs to know the cell size in order to compute
the coordinates for the display.

This is ment to clearly illustrate how the library is
working.  The intent is to be verbose in the relationship
that each value has relative to how it is used to affect
the resulting output.

The original code has too many lists of numbers to be
a good reference example.  By providing named colors
the code becomes easier to understand and reconcile with
the output displayed.

If you see silly variable names and casual underscores
here and there, it's because -Wshadow sucks in GCC 4.9.