Feature suggestion: Add extra hardware decoding devices

[Brainiarc7]

Suggestion: Add extra hardware decoding devices support (under the Hardware Optimizations page, Hardware Decoding Tab drop-down menu, with selectable devices per hardware acceleration setting basis.

Reason: On Multi-GPU systems, such as laptops with switchable graphics, you'll want to associate each hardware acceleration method, say, NVENC, to a DXVA context bound to the NVIDIA GPU, and another DXVA context bound to the Intel GPU for QuickSync acceleration on Optimus systems, and a similar setup for AMD VCE for AMD Enduro systems following the same syntax.

Viability: By default, on switchable graphics systems, running an application such as ConvertXtoHD will by default bind it to a given GPU depending on the default or user-set 3D profile on the system. This has a very detrimental effect on GPU - accelerated encoding performance on all fronts, requiring workarounds such as a user-set 3D profile to force the use of either the discrete or integrated GPU, and enabling only one of the acceleration options, and discarding the other.
As observed, the current situation is very suboptimal, resulting in degraded performance by default, and with user intervention, unused silicon die that could be put to use elsewhere.

This solution would also benefit PC users who do have systems where both the Intel Integrated Graphics and the discrete GPUs (AMD or NVIDIA with VCE and NVENC enabled respectively) can be put to use with greater overall performance.

Recommended fix: Create a test branch of ConvertXtoHD (and ConvertXtoVideo, as its' also affected) , make it available as a beta and the community will get to benefit from both the new features and in turn, providing usable feedback to the developers. I, for one, have hardware on which these changes could be tested on, and I volunteer for these tests.