Fix PrintHilbertSeries to also work in Python 2.x

[fingolfin]

It raises an exception right now:

Exception raised:
    Traceback (most recent call last):
      File "/opt/python/2.7.15/lib/python2.7/doctest.py", line 1315, in __run
        compileflags, 1) in test.globs
      File "<doctest rational.txt[6]>", line 1, in <module>
        cone.PrintHilbertSeries()
      File "/home/travis/build/Normaliz/PyNormaliz/PyNormaliz.py", line 242, in PrintHilbertSeries
        print(self.PrintPrettyHilbertSeries(numerator,denominator))
      File "/home/travis/build/Normaliz/PyNormaliz/PyNormaliz.py", line 231, in PrintPrettyHilbertSeries
        num, den = self.PrettyPolynomialTuple(numCoefficients, denCoefficients)
      File "/home/travis/build/Normaliz/PyNormaliz/PyNormaliz.py", line 197, in PrettyPolynomialTuple
        num = getNumerator(numCoefficients)
      File "/home/travis/build/Normaliz/PyNormaliz/PyNormaliz.py", line 186, in getNumerator
        coefficient) else '- ', abs(coefficient), to_sup(exp))
    UnicodeEncodeError: 'ascii' codec can't encode character u'\xb2' in position 0: ordinal not in range(128)

[w-bruns]

I think it is impossible to write the exponent as a superscript in Python2, at least it does not work as in the code.

Way out: make to_up dependent on the Python version (if possible) and use ^x in Python2.

[fingolfin]

I have a hard time believing that it would be impossible to do it in Python 2. It might require different code, and it might be very difficult to impossible to write code which works simultaneously in Python 2 and 3 (at least w/o sacrificing code readability seriously; after all, one could always write out byte arrays with the required byte sequences).

That said, I am also highly sceptical how useful this whole function is overall. It's a nice gimmick, but does it really belong into PyNormaliz? What about users whose terminal is not set to UTF (admittedly, that number might be small today, but I am just guessing here).

[w-bruns]

No doubt, it is kind of a gimmick, but the data in the raw output format are often difficult to read. And I think there could be more such functions.

We can of course replace the true exponents by numbers following ^. Then the problem is solved.

[w-bruns]

Now also available in Python 2 with ^ instead of proper superscript.