define octave_numeric_limits template methods for Inf and NaN * lo-ieee.h (octave_numeric_limits::Inf, octave_numeric_limits::NaN): New methods. * Canvas.cc, data.cc, graphics.cc, graphics.in.h, oct-handle.h, oct-stream.cc, quadcc.cc, sparse-xdiv.cc, str2double.cc, variables.h, __glpk__.cc, __init_fltk__.cc, __voronoi__.cc, octave.cc, Matrix.cc, CNDArray.cc, CSparse.cc, Sparse-C.cc, dDiagMatrix.cc, dMatrix.cc, dSparse.cc, fCMatrix.cc, fCNDArray.cc, fDiagMatrix.cc, fMatrix.cc, Faddeeva.cc, eigs-base.cc, lo-mappers.cc, lo-mappers.h, lo-specfun.cc, oct-norm.cc, oct-rand.cc, oct-rand.h, oct-spparms.cc, randgamma.cc, randpoisson.cc: Use them instead of octave_Inf, octave_Float_Inf, octave_NaN, and octave_Float_NaN.

gnulib-hg = gnulib-hg