Once this gets merged, we should release v0.1-alpha.4

• REBASE ONTO MAIN!

Major Changes

• Rewrite rendergraph and simplify it by removing polymorphism
• Implement support for dynamic rendering #438
• Multisampling with rendergraph #203
• Use VK_PRESENT_MODE_IMMEDIATE_KHR when vsync is turned off #530
• Refactor buffer and descriptor management into rendergraph #526
• Refactor descriptors and descriptor builder so it becomes a part of rendergraph #430
• ImGUI not being updated with new camera information when you move the camera just a little #405
• Modify descriptor builder to allow for multiple images to be passed into the builder pattern. #285
• Use builder pattern for the graphics pipeline #277
• Rendergraph input validation #204
• Add v0.1-alpha.4 changelog to wiki #312
• Add a reference page about the rendergraph to our docs
• Validation error: Hazard WRITE_AFTER_WRITE vs. layout transition #546
• Remove incorrect manual use of VK_LAYER_RENDERDOC_Capture! #547
• Use VK_EXT_debug_utils instead of VK_EXT_debug_report  #537
• Add wrapper for VkSampler
• Add wrapper for VkGraphicsPipeline
• Add wrapper for VkPipelineLayout
• Add more methods in make_info.cpp
• Remove unused msaa_target.hpp
• Use VK_API_VERSION_1_3 for Vulkan Memory Allocator
• Expose a simple API for Vulkan-renderer applications #297 (This would be too much for one pull request!)

Minor Changes

• Throw exception if name is empty in every RAII wrapper class
• Do not call .reset() explicitely before smart pointer reassignment (this is redundant)

Outlook (rendergraph3)

• Automatic double or triple buffering of all rendergraph resources to manage frames in flight automatically
• Parallelize Command Buffer recording
• Vulkan performance query pools
• Improve barrier placement
• Improve VkWriteDescriptorSet updating mechanism
• Pre-fill VkRenderingAttachmentInfo instead of doing it per-frame
• Internal descriptor performance metrics based on cmd_buf.bind_descriptor_set wrapper function
• SPIR-V shader metadata reflection by parsing with SPIRV-Cross
• Maybe rendergraph can reason about descriptor pool sizes ahead of descriptor pool allocation?
• Batch calls to vkCreateGraphicsPipelines (How to work this into GraphicsPipelineBuilder and ? (GraphicsPass wrapper?)
• Debatable: Implement a PipelineBarrierBatchBuilder? (Real use might not be significant)
• Check limits of VkPhysicalDevicewith respect to rendergraph setup?