macosx_vout: fixed flicker when resizing the vout through AR or crop changes as suggested by David Fuhrmann

also includes a minor fix to make room for the wordy German l10n

Close #3841

This updates the theora module to use the newer API introduced in Theora version 1.0.
It solves a crash on windows that was caused by reading past the end of the YUV structure due to an incorrect line count.
This has been fixed by using the minimum line count common to both the source and destination of the pixel data.
The new ycbcr structure in the new API makes the picture copy much more straightforward.
Modified-by: Jean-Baptiste Kempf <jb@videolan.org>
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

This requires Linux DVB 5.5.

It's the only output format on my GMA 3150
It is identical to YV12 except:
 - U & V are swapped
 - U & V pitch is identical to Y pitch (first half is plane data,
 second plane is padding)

Fixes #6008

Change --directory-version-sort boolean config option to a --directory-sort multiple choice option. This makes more sense and makes it possible to provide more sort options if needed without breaking backwards compatibility. The description strings are a bit long but they more or less fit in the prefs dialog here.

If you try to combine both features, you will end up in an fullscreen window with
playlist and overlayed video controls. Therefore make sure that both cannot be enabled
in simple prefs, and in case the user used advanced prefs, priorize nonembedded window
over the lion fullscreen.
Signed-off-by: Felix Paul Kühne <fkuehne@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

macosx: re-write the resize control used in the black window style to use the MouseDown event instead of the MouseDragged event which is ignored on Leopard (should fix #5822)

macosx: make sure the fspanel is visible when using 'float-on-top' in combination with Lion's native fullscreen mode

console and socket mode are now exclusive on windows.

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

When the code for detecting the number of CPUs present was moved
the headers necessary to compile the AltiVec detection code on
OpenBSD/powerpc was mistakenly removed so this puts the headers
back as appropriate.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

It seems to also be used for chroma upsampling.
It fixes #5751.

There are no reasons to only apply them to the first one. Certainly
some options might only make sense with a single item. But then it
really boils down to having only one item. Some options *do* make
sense for all (>1) items.