name2unicode(): handle hexadecimal literals for unicode glyphs in text extraction

[0xabu]

This is based on a slight tweak to the fix proposed by @janslifka in #183 (handling of lowercase hex literals, because those showed up in the sample PDF from issue #229).

Fixes #183, #229

[pietermarsman]

@0xabu, could you add a regression test that prevents the same mistake in the future?

[pietermarsman]

I am not an expert on this matter. But the Adobe Glyph List Specification does only talk about hexadecimal numbers, so this seems legit.

Also, two people indicate that this solves there problems.

So, I think we should merge this.

[0xabu]

@pietermarsman thanks for the pointer to the spec. That's very helpful to explain what is going on, but there are at least two issues:

• The spec makes it clear that "uni..." is a string of possibly multiple characters, whereas "u..." is always a single character. This code makes no attempt to handle that, in fact the use of re.search makes it look prone to mis-matching.
• The spec talks only about uppercase, but clearly PDFs in the wild use lowercase.

Nevertheless, I'm also tempted to merge this because it clearly makes things better than they were before.