This is yet another example based on digitised elevation contours. It provides
a particularly good torture test for the numerics involved in Sibson 
interpolation using Watson's method. In fact, it forced me to make changes to 
the code that handles degenerate cases during Sibson interpolation in v. 1.69.
Many thanks to Nick Cahill for identifying the problem and providing the data 
and to Maciej Sieczka for putting up the bug report.

To reproduce the example, run "./test.sh" or "make".
To visualize, in Matlab run "viewinterp('data.txt', 'nn.txt');".
To clean up, run "make clean".

Good luck!
Pavel Sakov
