buffer_cache: Misc fixups related to buffer clears

Improve management of kernel objects

vk_texture_cache: Use VK_IMAGE_LAYOUT_TRANSFER_SRC_OPTIMAL when possible

yuzu_cmd: Make use of fullscreen_mode setting

input_common: Make button threshold customizable

Reverts 48259de0c1a6a1aca77eec31cb8aca5ca2b680dd to the previous
hierarchy and fixes the resolution issue with this fullscreen mode.
yuzu-cmd will now read the fullscreen_mode setting and use it
appropriately.

- We're currently timing out with builds exceeding 60M. This doubles the timeout to 120M.

* emu_window_sdl2_vk: Use the generated SDL config

On Linux, due to the way we include SDL2 as a submodule, it makes it
difficult for us to specify which SDL_config.h we intended to include.
Before, CMake would default to the dummy one included with SDL and
ignore the generated one.

This tells CMake to use the generated one. In addition, we define
USING_GENERATED_CONFIG_H to throw an error in case the dummy config is
used by accident. Fixes Vulkan not working on Linux yuzu-cmd.

* emu_window_sdl2_vk: Specify the window manager if it should be supported

The original language "not implemented" is wrong if the implementation
exists but is not compiled. This causes a bit of a debugging headache
when it goes wrong. Log it if the window manager is known before
exiting.

* sdl_impl, emu_window: Remove clang ignore

Fixed upstream by
libsdl-org/SDL@25fc40b0bd44c484051064bc6b945ea9943f88dd

* Enable fullscreen support for Vulkan on yuzu-cmd

Hooked up the existing SDL2 logic for fullscreen support in the Vulkan window of yuzu-cmd.

* Change fullscreen logic to attempt desktop resolution first on yuzu-cmd

Changed the order in which we attempt to switch to fullscreen. First try desktop resolution first, if it fails fall back to streched fullscreen using windowed resolution.

Co-authored-by: lat9nq <22451773+lat9nq@users.noreply.github.com>
Co-authored-by: san <san+gitkraken@smederijmerlijn.nl>

applet_controller: Add preliminary support for version 8

- ExitProcess is not actually implemented either way, and this needs more work before we implement.

- Since we do not implement multiprocess right now, this should not be a crashing assert.

input_common: Support SDL toggle buttons

Avoids reading out of bounds from the stack.

Use existing helper functions and avoid looping when
only one buffer has to be active.

DMAEngine: Accelerate BufferClear [accelerateDMA Part 2]

bootmanager: Create a dummy render widget

gl_texture_cache: Workaround slow PBO downloads on radeonsi

uuid: Directly compare UUID instead of checking per element

Silences performance warnings generated from validation layers on each frame.

There's an optimization bug on non-git mesa versions where not
specifying GL_CLIENT_STORAGE_BIT causes very slow reads on the CPU
side.

Add this bit for all vendors.

We can now update this for C++20

- value return can be different 0, is not error is normal, error is only -1.

Version 8 adds support for key remapping introduced in FW 11.0, we will not be implementing this for now.

This ensures that Qt positions the render window at the correct position on initializing the respective render backends.

Adds support for loading games with multiple programs embedded within such as the Dragon Quest 1+2+3 Collection