handle xarrs with quantities in places they were raising issues

[keflavich]

Glen Langston reported that this example:
https://pyspeckit.readthedocs.io/en/latest/example_continuum_fromscratch.html
was failing with the following traceback:

Traceback (most recent call last):
  File "<ipython-input-1-dca5499ff34c>", line 29, in <module>
    sp.specfit(fittype='polycontinuum', guesses=(0,0), exclude=[30, 70])
  File "/home/adam/repos/pyspeckit/pyspeckit/config.py", line 145, in decorator
    f(self, *args, **new_kwargs)
  File "/home/adam/repos/pyspeckit/pyspeckit/spectrum/fitters.py", line 344, in __call__
    self.multifit(show_components=show_components, verbose=verbose,
  File "/home/adam/repos/pyspeckit/pyspeckit/spectrum/fitters.py", line 822, in multifit
    self.plot_fit(annotate=annotate,
  File "/home/adam/repos/pyspeckit/pyspeckit/spectrum/fitters.py", line 1145, in plot_fit
    self._full_model()
  File "/home/adam/repos/pyspeckit/pyspeckit/spectrum/fitters.py", line 1035, in _full_model
    self.fullmodel = self.get_full_model(debug=debug,**kwargs)
  File "/home/adam/repos/pyspeckit/pyspeckit/spectrum/fitters.py", line 1040, in get_full_model
    return self.get_model(self.Spectrum.xarr, debug=debug,**kwargs)
  File "/home/adam/repos/pyspeckit/pyspeckit/spectrum/fitters.py", line 1045, in get_model
    return self.get_model_frompars(xarr=xarr, pars=self.parinfo,
  File "/home/adam/repos/pyspeckit/pyspeckit/spectrum/fitters.py", line 1057, in get_model_frompars
    return (self.fitter.n_modelfunc(pars,
  File "/home/adam/repos/pyspeckit/pyspeckit/spectrum/models/model.py", line 402, in L
    v += self.modelfunc(x, *parvals[lower_parind:upper_parind], **kwargs)
  File "/home/adam/repos/pyspeckit/pyspeckit/spectrum/models/polynomial_continuum.py", line 24, in polymodel
    return numpy.polyval(pars,x)
  File "<__array_function__ internals>", line 180, in polyval
TypeError: no implementation found for 'numpy.polyval' on types that implement __array_function__: [<class 'pyspeckit.spectrum.units.SpectroscopicAxis'>]

This is caused by some change in how Quantities behave wrt numpy functions. This PR implements some workarounds that strip quantity properties where they might cause problems.