DOLT (provides doltlibtool and doltcompile to make compilation faster
by not always running full libtool) does not handle --with-pic at all,
so even when PIC is requested, it won't use it.

As a workaround manually add it to X86ASMDEFS.

This reverts commit 508e3db7.

This fixes meson build issues when cross-compiling for iOS.

The libtool in extras/tools has additional patches that are needed to
build with llvm-mingw.

We need to keep a local copy of the pool configuration now.

+ rename DestroyVideoDecoder to ReleasePoolSurfaces

Only D3D11 was using it and it doesn't make sense in push.

va_surface is always set in picture context

reindent after external pool removal

We can push pictures to the display with zero copy as this is the same device.

vout_device_configuration_t is for device configuration
vout_configuration_t is for display configuration

We don't need a full vout_configuration_t for that, only the vout object and the
video_format_t to initialize the window size.

instead of the decoder device

We might store a reference to the VA module so the destructor of the module is
not called while the video context is used.

For now they don't create one.

Each video context can store extra objects, like the IDirect3DDevice9* for D3D9,
which is not found in the decoder device.

It's kept in the input resources and will come embedded in the video context.

Decoders that don't use a decoder device don't need to create one in the first
place.

The pictures are similar to the ones created in the OpenGL converter.

And remove the locally held decoder device that used to come from the
picture_sys_t.

rather than the picture_sys

rather than the picture_sys

This is used to initialize the VA instead of the picture_sys_t from the
external decoder pool.

Only call decoder_GetDecoderDevice() in lavc_UpdateVideoFormat(). The
decoder_UpdateVideoOutput() is done afterwards.

After the call to lavc_UpdateVideoFormat() the received decoder device holds an
extra reference that needs to be released after the VA is created.

decoder_NewPicture() cannot be used until decoder_UpdateVideoOutput() is called.
So this patch temporarily breaks va modules relying on the picture_sys_t data
that used to be passed. It will be via the decoder device we just got.

This is how the VA's will be probed when multiple outputs are possible.

The video context will be passed to various modules up to the display where it
will be ultimately released when the last associated picture will be released.
It needs to be refcounted and we need a destructor to do this final release.

By default the video context holds a reference to the matching decoder device.
This reference is released with the video context.

Rather than telling the resource storage if the vout needs to be stopped in some way.

Incidentally the vout_Cancel is not needed anymore since the vout_StopDisplay, which
is now used, destroys the picture pool.

First handle the resources needed by the get_device() call, then the
format_update() which creates the display.

The on_vout_started is only notified when the vout is really started.

Rename vout_thread_added to vout_thread_started.

It may not work, in which case a new display/thread should be created to handle
the new format.

No functional changes.

This is the real event the player is looking for. It doesn't care if a vout has
been created but is dormant.

No functional changes.

To create the vout and decoder device we don't need to know the DPB.

For now the display still holds the picture pool so the display still needs the
DBP size.