add iteration limit to the private fitGAM function to avoid infinite …

…loop with Netlib BLAS/LAPACK, issue #160

inst/Regression/RegressionGAM.m

@@ -715,8 +715,9 @@ function savemodel (obj, fname)
       ns        = rows (X);
       Tol       = this.Tol;
       ## Start training
-      while (! converged)
+      while (! converged || iter > 1000)
         iter += 1;
+
         ## Single cycle of backfit
         for j = 1:columns (X)