get_physical_device_surface_capabilities2 should take a mutable reference to vk::SurfaceCapabilities2KHR

[JoshuaMasci]

The function vkGetPhysicalDeviceSurfaceCapabilities2KHR can be extended by use of VkSurfaceCapabilitiesKHR.pNext, however the current wrapper get_physical_device_surface_capabilities2 doesn't allow this since it creates and returns the struct vk::SurfaceCapabilities2KHR.

To allow the use of the pNext pointer, the function get_physical_device_surface_capabilities2 should be changed from this:

  pub unsafe fn get_physical_device_surface_capabilities2(
      &self,
      physical_device: vk::PhysicalDevice,
      surface_info: &vk::PhysicalDeviceSurfaceInfo2KHR,
  ) -> VkResult<vk::SurfaceCapabilities2KHR>

To something like this:

  pub unsafe fn get_physical_device_surface_capabilities2(
      &self,
      physical_device: vk::PhysicalDevice,
      surface_info: &vk::PhysicalDeviceSurfaceInfo2KHR,
      surface_capabilities: &mut vk::SurfaceCapabilities2KHR,
  ) -> VkResult<()>

[MarijnS95]

We've had discussions about how to best pass these types in the past and fixed most functions to allow this (e.g. #588), but it seems like it was forgotten about when this extension was originally wrapped in #530. Especially considering that VkSurfaceCapabilities2KHR exists solely to represent VkSurfaceCapabilitiesKHR but with sType and pNext fields 😅.

It looks like at least VkSurfacePresentModeEXT is relying on this behaviour currently.