[tex-hyphen] Changes in collection-lang*

[Karl Berry]

    - Turkmen is not an European language.
    - Georgian is not written in Cyrillic script.

Right, thanks, fixed x 2.

    - (I would also say that "latin serbian" is not written in Cyrillic,
    but that would only make it *really* confusing to have serbian split
    into "Cyrillic" and "European" part, 

I came to the same conclusion.

    - Mongolian: (may be) written in Cyrillic script and hyphenation
    patterns are for Cyrillic - how does one know where it belongs?

If you think "cyrillic" is preferable, I'll move it.  I had the
imprression more of the Mongolian stuff we had was about non-Cyrillic,
but do I know Mongolian?  No :).

    - Afrikaans: how does one know that it belongs to 'other' and not to
      'african'?

1) Because it's written in Latin and hence is not an African "script",
which is what langafrican was created for (not by me).  Not that any
such distinctions are rigorously followed.  I guess by analogy with the
langeuro that I just created (not without much trepidation, I assure
you), perhaps it would be better to move it. 

2) "How does one know" is always the wrong question to ask.  In practice,
there is no reliable way to predict in which collection support for a
given language will reside.  If one really wants to know, one just has
to look.  If you think there is some perfect slicing and dicing of all
this that will meet all user and installer expectations, let me know :).

My basic theory is that it doesn't matter (fortunately), because nearly
all users can just install everything.

    - It makes sense to merge, say, German documentation into
    collection-langgerman, but I find it a bit confusing that
    collection-langenglish now contains a lot of documentation that
    might have nothing to do with support for English (for example
    xetexref).

Indeed, "support for English" has nothing to do with it.  Just as, say,
microtype-de or texlive-de has nothing to do with "support for German".

The basic idea of the language collection is that if you don't speak,
say, German, you could in principle omit the collection and not miss
anything.  So, I figured English was the same.  If one can't read
English at some level, xetexref doesn't do you any good.  

That said, I was unsure about moving all these too.  I admit it's
tempting to put xetexref into xetex.  But, well, I was getting rid of
all c-documentation-* and so I had to put them somewhere, and this was
the obvious first stop.  I am open to arguments for moving stuff
around.  I never thought this first cut would also be the last word.

karl