75 characters is the maximum length of a line before gettext breaks it,
so this convention will cause each translatable string to fit snugly
into the po files. It's also about what we were doing anyway, just not
consistently.

The biggest advantage to consistent line wrapping is that translators
will no longer have to guess where they should insert line breaks.

I have rewrapped the Catalan, Czech, Danish, German, Spanish, Finnish,
French, Croatian, Hungarian, Italian, Lithuanian, Norwegian, Dutch,
Polish, Portuguese, Romanian, Russian, Slovene, Swedish, Turkish, and
Ukranian translations so as to not burden their maintainers.

The Bulgarian, Greek, Esperanto, Farsi, Hindi, Malayalam, Oriya, Punjabi,
Romansh, Slovak, Serbian, Telugu, Thai, and Walloon translations did not
need to be rewrapped.

I removed Hebrew "translations" of cmd.rc that were just copies of the
English strings, and marked cmd.rc:142 as fuzzy if the translation
mentioned wine.conf.

Single-line translations that were 80 characters long or less were not
wrapped.

The Arabic, Japanese, Korean, and Chinese translations still need to be
rewrapped.

When the number of elements per bucket is too small, then try to reduce
the number of threads by merging buckets. This is to ensure that the
number of running wait queue threads doesn't get too big.

To implement waiting for an arbitrary number of handles, we group them
in buckets up to (MAXIMUM_WAIT_OBJECTS - 1) objects, and then assign a
dedicated wait queue thread. The last handle is used to notify about
changes.

Also lock/unlock the module only on creation/destruction of the object.

Also lock/unlock the module only on creation/destruction of the object.

Also lock/unlock the module only on creation/destruction of the object.

Also lock/unlock the module only on creation/destruction of the object.

Also lock/unlock the module only on creation/destruction of the object.