Xenoblade 2 invokes a draw call with zero vertices.
This is likely due to indirect drawing (glDrawArraysIndirect).

This causes a crash in the staging buffer pool when trying to create a
buffer with a size of zero. To workaround this, skip index buffer setup
entirely when the number of indices is zero.

vk_renderpass_cache: Pack renderpass cache key and unify keys

vk_memory_manager: Remove unified memory model flag

services: hid: Stub StopSevenSixAxisSensor.

Optimize GPU Command Lists and Introduce Fast GPU Time Option

Fix "Port citra-emu/citra#4956: "Fixes to game list sorting" #3611"

service: Update function tables

{gl,vk}_rasterizer: Add lazy default buffer maker and use it for empty buffers

configuration: Add Restore Default and Clear options to hotkeys

configure_input_player: Use slider to edit modifier scale

- Used by The Legend of Zelda: Breath of the Wild v1.6.0

vulkan/wrapper: Sort physical devices

decode/arithmetic_half: Fix HADD2 and HMUL2 absolute and negation bits

svc: Re-add MapProcessCodeMemory/UnmapProcessCodeMemory

Sometimes for unknown reasons NVN games can bind a render target format
of 0. This may be a yuzu bug.

With the commits before this the formats were specified without being
"packed", assuming all formats and texceptions will be written like in
the color_attachments vector.

To address this issue, iterate all render targets and pack them as they
are valid. This way they will match color_attachments.

- Fixes validation errors and graphical issues on Breath of the Wild.

shader/arithmetic_integer: Fix LEA_IMM encoding

Fix -Werror=conversion and -Wdeprecated-copy issues

frontend/filesystem: Add a trailing separator to the string path

page_table: Remove unused captures

svc: Remove unused variable

shared_memory: Amend doxygen reference

Amends the parameter to match the documentation reference.

Resolves a -Wdocumentation warning with clang.

These were lost in the re-implementation of the virtual memory manager.

Since the VMM refactor, this is no longer used or needed.

Any time the lambda function is called, the permission being used in the
capture would be passed in as an argument to the lambda, so the capture
is unnecessary.

The encoding for negation and absolute value was wrong.
Extracting is now done manually. Similar instructions having different
encodings is the rule, not the exception. To keep sanity and readability
I preferred to extract the desired bit manually.

This is implemented against nxas:
https://github.com/ReinUsesLisp/nxas/blob/8dbc38995711cc12206aa370145a3a02665fd989/table.h#L68

That is itself tested against nvdisasm (Nvidia's official disassembler).

Framebuffer creation was ignoring the number of color attachments.

This allows us to call Common::CityHash and std::memcmp only once for
GraphicsPipelineCacheKey. While we are at it, do the same for compute.

Fix format error in performance statistics

* kernel: memory: Improve implementation of device shared memory.

* fixup! kernel: memory: Improve implementation of device shared memory.

* fixup! kernel: memory: Improve implementation of device shared memory.

service/time: Remove reliance on the global system accessor