RayTracing Invocation Reorder

First public implementation (more clear and correct than official Vulkan docs of the extension) of SPV_KHR_shader_invocation_reorder in HLSL, see our spirv_intrinsics headers in nbl/builtin/hlsl

Solves: microsoft/DirectXShaderCompiler#8226

What's Changed

New Features

• Nabla can be consumed as a CMake package and has a target to make a redistributable SDK
• CI now builds Nabla redistributables for Win32 which can be downloaded
• Nabla Shader Compiler Boost.Wave preprocessor reworked, has a cache and extremely verbose runtime error logging
• Some examples now compile runtime shaders in parallel and use pipeline cache
• Mitsuba Scene Loader resurrected
• Material Compiler v3 Frontend AST complete
• Monte Carlo Sampling HLSL library with unit tests and benchmarks
• CDF HLSL sampler
• IES profile loader and viewer example
• Alias Table Builder and HLSL sampler
• Resampled Importance Sampling compatible BxDFs and Light Samplers
• Statically Polymorphic HLSL Path Tracer
• Example 40 a WIP Path Tracer
• SER / Shader Invocation Reorder support
• OBB calculation for Polygon Geometries
• Vertex Welding for Polygon Geometries
• Iridescent Fresnel (thin two-layer diffractive) implementation for BRDFs and BSDFs
• Statically Polymorphic adaptive Simpson Quadrature in HLSL
• octahedral mapping HLSL header
• Cache for Owen Scrambled Sobol Sequence generation

Improvements

• Far more performant and correct Spherical Triangle (Arvo 1997) and Spherical Rectangle (Urena 2013) sampling on the GPU
• Experimental DXC and SPIR-V Tools legalization flags which reduce compile times by a factor of 10 via limiting unrolling before legalization
• refactored the material_compiler3/CNodePool to have its functionality in CObjectPool
• New Curve fit fast approximations for acos(x)/sin(x)
• DXC and SPIR-V Tools Updates
• experimental and draft ICPUScene resource
• add getGeometryExclusiveCount to IMorphTargets
• IRayTracingPipeline SBT validation
• add interm_getImageViewInHierarchy to IAssetLoader
• HLSL binary searches can now take stateful comparators (mostly to track accesses for CDF sampler)
• reworked concepts/accessors for images
• add some scalar-component-wise matrix operators
• Owen Scrambled Sobol Sequence generation can be parallelized now

Bugfixes

• bugs in BLAS validation
• Commandbuffer tracking of resources used by a command if more than 32k shared pointers referenced
• Smooth Normal recomputation for Polygon Geometries is now correct
• Cleaner Example shutdowns in event of a failure
• ICPUGeometryCollection::valid logic
• rework exportForBLAS method into a CBLASExporter and IBLASExporter classes respectively for ICPUGeometryCollection and IGeometryCollection respectively

Deprecations / removals

• old matrix4SIMD and matrix3x4SIMD classes
• old core::quaternion now HLSL version only used
• SStridedRange<BufferType> moved to core namespace
• IBottomLevelAccelerationStructure::BUILD_FLAGS::ALLOW_DATA_ACCESS_BIT removed _BIT from enum name
• ICPUTopLevelAccelerationStructure::getBuildRangeInfo removed
• morton code HLSL and C++ reworked and unified

Full Changelog: v0.9.1.0...v0.10.0.0