Fix performance regression introduced in godotengine#90993

PR godotengine#90993 needed to get rid of VMA_MEMORY_USAGE_AUTO_PREFER_HOST because
we no longer used vmaCreateBuffer so we could specify the allocation
callbacks.

This however resulted in the wrong memory pool being chosen, causing
signficant performance slowdown.

Indicate additional preferred flags to help VMA select the proper pool.

Fixes godotengine#101905

Author: darksylinc

drivers/vulkan/rendering_device_driver_vulkan.cpp

@@ -1530,10 +1530,12 @@ RDD::BufferID RenderingDeviceDriverVulkan::buffer_create(uint64_t p_size, BitFie
 			if (is_src && !is_dst) {
 				// Looks like a staging buffer: CPU maps, writes sequentially, then GPU copies to VRAM.
 				alloc_create_info.flags = VMA_ALLOCATION_CREATE_HOST_ACCESS_SEQUENTIAL_WRITE_BIT;
+				alloc_create_info.preferredFlags |= VK_MEMORY_PROPERTY_DEVICE_LOCAL_BIT;
 			}
 			if (is_dst && !is_src) {
 				// Looks like a readback buffer: GPU copies from VRAM, then CPU maps and reads.
 				alloc_create_info.flags = VMA_ALLOCATION_CREATE_HOST_ACCESS_RANDOM_BIT;
+				alloc_create_info.preferredFlags |= VK_MEMORY_PROPERTY_HOST_CACHED_BIT;
 			}
 			alloc_create_info.requiredFlags = (VK_MEMORY_PROPERTY_HOST_COHERENT_BIT | VK_MEMORY_PROPERTY_HOST_VISIBLE_BIT);
 		} break;