coreinit: Added support for 'wb' modifier flag in FSOpenFile()/SAVEOpenFile()

AX: Rewrote a majority of AX API and internal mixing code, aiming for almost a 1:1 reimplementation where possible
AX: Adjusted memory layout of some internal sample data buffers (some games calculate addresses on their own instead of relying on the values provided by the API)

SWKBD: Fixed a bug that caused the software keyboard overlay to not display on AMD GPUs

GX2: Added support for D24_S8_FLOAT depth buffer format
GX2: Shaders will now correctly use shadow samplers instead of imitating them by manually comparing shadow depth in shader logic
GX2: Added support for dynamically detecting and adjusting the number of mip levels of a texture during usage (information about the mip level count isn't always available at texture creation)
GX2: Better support for rendering to individual mip levels of a texture
GX2: Added support for binding the same texture multiple times with different mip or slice views during a drawcall
GX2: Fixed order of elements returned by FETCH4 shader instruction
GX2: Fixed a bug in polygon offset slope scale calculation
GX2: Handle texture LOD parameter correctly in vertex shader

# CEMU 1.6.2c

CPU: Fixed a bug in recompiler PSQ_STU instruction

# CEMU 1.6.2 + 1.6.2b

CPU: Fixed a rare crash bug that could occur when the recompiler attempted to analyze certain functions
CPU: Improved detection of function entry points for recompiler
CPU: Added instructions LBZUX, SRW., PSQ_LU, PSQ_STU and PS_DIV to recompiler
CPU: Adjusted recompiler PS_MERGE10 instruction to generate faster code

FS: Various optimizations to FSA thread, resulting in quicker processing of file operations
FS: If available, AES-NI is now used for filesystem decryption

AX: Added API AXAcquireMultiVoice()

NFP: Deactivate event will now be correctly signaled

GX2: Added support for Streamout emulation via OpenGL's Transform Feedback
GX2: Slightly reduced the number of shaders that need to be compiled (by optimizing away render states that previously affected the generated shader code)
GX2: Improved handling of bottom of pipe event
GX2: Added API GX2DrawIndexedImmediateEx()
GX2: Fixed a bug in the buffer cache that could cause out-of-bounds allocations which would result in CEMU crashing
GX2: Fixed a bug where textures would be immediately reloaded from RAM after being updated by a GPU-side operation

general: Fixed a bug where files would be created in/loaded from the wrong directory after the load file dialog.
NFP: Added initial NFC support
	 Scanning NFC tags works by loading a file via the menu
GX2: New universal cache implementation for attribute, uniform and index data
     Additionally, there is now a cache accuracy option to control the level of accuracy vs performance. Can be changed while a game is already running. 
	 The previously added cache control options where removed as they have become obsolete 
GX2: Further improved and optimized handling of textures with overlapping memory regions
GX2: Optimized detection of changes to texture data
GX2: Adjusted the idle loop of the GPU7 command processor thread to respond quicker to new data
GX2: Fixed a crash that could occur if GL shaders failed to compile
GX2: Moved handling of alpha test into shader code (rather than relying on OpenGL's deprecated hw alpha test)
GX2: Fixed a bug where the swizzle value of a texture would not be correctly updated when an already cached texture was reloaded