Hungarian dev set released, minor fix to Czech news data

[Philipp Koehn]

Hi,

two minor updates for the WMT 2009 shared task:

* the development sets (which are cleaned-up versions of last year's test set)
now include the Hungarian sets.

* the monolingual news corpus was slightly corrected for Czech - the previous
release included some Slovak data

-phi