[audio_core] Use different state offset for each biquad filter channel

main: Fix borderless fullscreen for high dpi scaled displays

settings: added missing graphical settings to RestoreGlobalState()

On Windows, a borderless window will be treated the same as exclusive fullscreen
when the window geometry matches the physical dimensions of the screen.
However, with High DPI scaling, when the devicePixelRatioF() is > 1, the borderless
window apparently is not treated as exclusive fullscreen and functions correctly.
One can verify and replicate this behavior by using a high resolution (4K) display,
and switching between 100% and 200% scaling in Windows' display settings.
At 100%, without the addition of 1, it is treated as exclusive fullscreen.
At 200%, with or without the addition of 1, it is treated as borderless windowed.
Therefore, we can use (read: abuse) this difference in behavior to fix this issue for
those with higher resolution displays when the Qt scaling ratio is > 1.
Should this behavior be changed in the future, please revisit this workaround.

core: hid: Use gyro thresholds modes set by the game

texture_cache: OpenGL: Implement MSAA uploads and copies

added missing graphical settings to RestoreGlobalState()

kernel/svc: Fix undefined info_id

video_core: Speed up video frame data copy

core: kernel: k_process: Use application system resource.

biquad_filter: Fix rounding in ApplyBiquadFilterInt

gpu_thread: Remove OnCommandListEndCommand

kernel/svc: switch to generated wrappers

input_common: Reintroduce custom pro controller support

main: Re-add QtWebEngine zoom factor

kernel: avoid usage of bit_cast

For some reason, I had removed this in https://github.com/yuzu-emu/yuzu/pull/4949/commits/ad6cec71ecd61aa2533d9efa89b68837516f8464

This should fix any improperly scaled web applets.

Remove fake vertex bindings when dynamic state is enabled

glsl_emit_context: Remove redeclarations of gl_SampleID and gl_SampleMask

audio: cubeb: Fix yuzu crashing when it test for latency

buffer_base: Partially revert changes from #9559

This fixes a regression where Yoshi's Crafted World (and potentially other titles) would enter an infinite loop when GPU Accuracy was set to "Normal"

These built-ins seem to be available without needing to be declared for fragment shaders, similar i.e. to gl_FragDepth

service: hid: Return error if arguments of SetSupportedNpadIdType is invalid

kernel: fix compilation with older gcc

Call rasterizer->ReleaseFences() directly