Basic analysis of the script content of strings

[jwiggins]

This is another part of #762

There are two pieces here, and a ton of machine-generated code (don't fear the diff):

• A small program which fetches http://www.unicode.org/Public/UNIDATA/Scripts.txt, parses it, and writes out the kiva.fonttools.text._data module.
• A new class UnicodeAnalyzer which uses the data from Scripts.txt and returns the languages and slices for a given input string.

UnicodeAnalyzer is pretty basic right now. I'd like to keep it that way in this PR. For instance, Emoji ligatures are not great:

In [3]: s = "👩‍👩‍👧‍👧" 

In [4]: s                                                                       
Out[4]: '👩\u200d👩\u200d👧\u200d👧'

In [5]: an.languages(s)                                                         
Out[5]: 
[(0, 1, 'Common'),
 (1, 2, 'Inherited'),
 (2, 3, 'Common'),
 (3, 4, 'Inherited'),
 (4, 5, 'Common'),
 (5, 6, 'Inherited'),
 (6, 7, 'Common')]

[jwiggins]

These were changed to match the names in kiva.fonttools.text._data.

[rahulporuri]

LGTM

[rahulporuri]

This looks like the most important detail in the PR - how we're selecting the entry given a code point - and it'd be useful if you could elaborate on how we're doing it.

[jwiggins]

Fair enough

[rahulporuri]

still LGTM

[jwiggins]

Thanks for the review