hasvk: Track DPB slots instead of assuming a single reference.

The encoder ignored VkVideoReferenceSlotInfoKHR entirely: it took
pReferenceSlots[0] as the one reference, wrote its address into frame store 0
and hardcoded the L0 list to that store.  Two DPB slots were all it could
honestly advertise, which is fewer than a GOP of any length needs.

Record each slot the operation names under its own index, since a slot maps
onto the frame store of the same index and MFX_PIPE_BUF_ADDR_STATE holds 16
of those, and build the L0 entries from StdVideoEncodeH264ReferenceListsInfo,
whose RefPicList0 already carries slot indices.  Clients that pass reference
slots without a list keep the old behaviour.

maxDpbSlots therefore goes to 16.  maxActiveReferencePictures stays at 1:
every macroblock is still coded against reference index 0, so a picture may
be kept in the DPB but only the one it names first is predicted from.

dEQP-VK.video.encode.h264.intra_refresh_* goes from 12 Pass and 32 stopped at
the maxDpbSlots >= GOP frame count precondition to 36 Pass, the remaining 8
being per-picture-partition mode, which is not offered.  The family as a whole
is 76 of 120 with no failures.  ffmpeg's h264_vulkan encoder no longer trips
its slot_index assertion and produces a full GOP.

Co-Authored-By: Claude Opus 5 <[email protected]>

src/intel/vulkan_hasvk/anv_video.c

@@ -202,11 +202,12 @@ anv_GetPhysicalDeviceVideoCapabilitiesKHR(VkPhysicalDevice physicalDevice,
       pCapabilities->stdHeaderVersion.specVersion =
          VK_STD_VULKAN_VIDEO_CODEC_H264_DECODE_SPEC_VERSION;
    } else {
-      /* Limits reflecting what the encoder emits: I and P frames, CQP, single
-       * reference. P frames use one reference (the previous reconstructed
-       * frame) plus the current setup slot, so two DPB slots and one active
-       * reference. B frames and multi-reference are not implemented. */
-      pCapabilities->maxDpbSlots = 2;
+      /* Limits reflecting what the encoder emits: I and P frames, CQP.  A DPB
+       * slot maps onto a frame store of the same index and there are 16 of
+       * those, so a long GOP can keep every picture it likes around.  Only one
+       * of them can be predicted from, though: each macroblock is coded against
+       * reference index 0, so multi-reference and B frames are still out. */
+      pCapabilities->maxDpbSlots = ANV_VIDEO_MAX_DPB_SLOTS;
       pCapabilities->maxActiveReferencePictures = 1;
 
       struct VkVideoEncodeCapabilitiesKHR *enc_caps =