Allow reading HTML tables with unicode values

[taldcroft]

Fixes #5409.

[saimn]

To avoid this and using TEST_UNICODE_LITERALS, the strings below could simply be defined with the u'...' syntax, or maybe I miss something ?

[taldcroft]

@saimn - good point. During development I did try using unicode strings at one point and this failed, but this might have been when there were other problems. I tried just and it worked as expected, so this is definitely cleaner. New rebased version pushed.

[mhvk]

Should this be a bug fix?

Also, I noticed this overrides default_converters from TableOutputter -- just in case: would it make sense to just change the default there? Or do other things go wrong then?

[taldcroft]

@mhvk - reading unicode in Python 2.7 is not supported per the following Note in the docs:

Reading a table which contains non-ASCII (unicode) characters is only supported in Python 3 or
greater. If you have Python 2.x and need this functionality, consider using a newer version of
Python.

So this is really not a bug because reading unicode is explicitly not supported. But of course we could be more liberal about the definition of a bug.

[taldcroft]

And as a reminder, unicode is not supported for Python 2.7 because all the other Python readers use the stdlib csv module for splitting text, and this does not support unicode without shenanigans. This falls in the category of "we're not going to spend effort making this work on Python 2 when you should instead migrate to Python 3."

[mhvk]

@taldcroft - OK, makes sense. I agree that we should not put effort in making this work on python2