Fix for failing test GaudiGSL.InterpTest

As far as I can tell, this test confuses interpolation and extrapolation. These have historically been the same in GSL, but haven't been for a while now. If you want to extrapolate, you have to tell GSL to do so explicitly.

Relevant GSL commit from 2010 which makes trying to extrapolate using interpolation routines an error : http://bzr.savannah.gnu.org/lh/gsl/trunk/revision/4704 .

I would thus propose modifying the test to solely do interpolation instead.

GaudiGSL/src/Tests/InterpTest.cpp

@@ -28,7 +28,7 @@ int main()
std::vector<double> vctx ;
std::vector<double> vcty ;
-  for ( int i = 0 ; i < 20 ; ++i )
+  for ( int i = 0 ; i <= 20 ; ++i )
{
double x = i * 3.141592/20.0 ;
double y = mysin ( x ) ;
@@ -55,7 +55,7 @@ int main()
const GaudiMath::Function& i4 = plusOne +
GaudiMath::SplineInteg(vctx, vcty, Cspline_Periodic, 0.0) * minusOne;
-  for ( int i = -10 ; i < 30 ; ++i ) {
+  for ( int i = 0 ; i < 30 ; ++i ) {
double x = i * 3.141592/30.0 ;
printf( "x=%+7.1f; Sin-i1=%+14.9f; Cos-i2=%+14.9f; Sin-i3=%+14.9f; Cos-i4=%+14.9f;\n",
x / Gaudi::Units::degree,