Bug: Throws NumOutOfRangeError for numbers longer than 36 digits

[torond]

A NumOutOfRangeError is thrown when numbers longer than 36 digits are converted.
This makes the system prone to break, when the input text is unknown.

To reproduce:

>>> from g2p_en import G2p
>>> g2p_en = G2p()
>>> g2p_en('999999999999999999999999999999999999')  # This works
['N', 'AY1', 'N', ' ', 'HH', 'AH1', 'N', ...]
>>> g2p_en('1000000000000000000000000000000000000')  # This doesn't
...
inflect.NumOutOfRangeError

This error stems from the inspect.py module, which isn't part of this projects source code.
Workaround: Remove numbers longer than 36 digits instead of transforming them to words. This would lose some edge case information but is the only way to keep the transcribed output clean.

[divyanx]

I also encountered this problem, just figured out a walk around.
For numbers bigger than those supported can just make to pronounce each digit instead of the whole place value system. This will be a bit in hand with what we do with large numbers.

[torond]

That would also be a justified solution.
Although it might add a lot of information and increase the noise of the output.
Especially if the author did not intend the numbers to be read out loud.

I don't recall how floating point numbers are transcribed (e.g. 0.2348719). They should definitely be transcribed digit-by-digit.