Fix mathops.h to build with cuda 12.

[okanaganrusty]

Patch forward referencing https://codereview.qt-project.org/c/qt/qtwebengine-chromium/+/509219/2/chromium/third_party/ffmpeg/libavcodec/x86/mathops.h to allow livepeer with CUDA 12 to build.

This fixes the following build time issue at commit 2e18d06.

$ make
CC	libavdevice/avdevice.o
AR	libavdevice/libavdevice.a
CC	libavfilter/avfilter.o
CC	libavfilter/vf_signature.o
./libavcodec/x86/mathops.h: Assembler messages:
./libavcodec/x86/mathops.h:125: Error: operand type mismatch for `shr'
...

CUDA version:

$ /usr/local/cuda/bin/nvcc -V
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2023 NVIDIA Corporation
Built on Wed_Nov_22_10:17:15_PST_2023
Cuda compilation tools, release 12.3, V12.3.107
Build cuda_12.3.r12.3/compiler.33567101_0

Linked against CUDA 12 shared libraries:

$ ldd ./livepeer | grep cuda
	libnppig.so.12 => /usr/local/cuda/targets/x86_64-linux/lib/libnppig.so.12 (0x00007d4fa2200000)
	libnppicc.so.12 => /usr/local/cuda/targets/x86_64-linux/lib/libnppicc.so.12 (0x00007d4fa1800000)
	libnppc.so.12 => /usr/local/cuda/targets/x86_64-linux/lib/libnppc.so.12 (0x00007d4fa1000000)