Crash when submitting two command buffers together that render to a surface

[HeroesGrave]

Description
If you create 2 command encoders and create a render pass in each referring to a surface (possibly can even be different surfaces but I haven't checked), then submit the resulting command buffers together, the program crashes

Repro steps
Make the following change to the cube example

diff --git a/wgpu/examples/cube/main.rs b/wgpu/examples/cube/main.rs
index dadf81b0..00f4b144 100644
--- a/wgpu/examples/cube/main.rs
+++ b/wgpu/examples/cube/main.rs
@@ -392,8 +392,24 @@ impl framework::Example for Example {
                 rpass.draw_indexed(0..self.index_count as u32, 0, 0..1);
             }
         }
-
-        queue.submit(Some(encoder.finish()));
+        let mut encoder2 =
+            device.create_command_encoder(&wgpu::CommandEncoderDescriptor { label: None });
+        {
+            let mut rpass = encoder2.begin_render_pass(&wgpu::RenderPassDescriptor {
+                label: None,
+                color_attachments: &[Some(wgpu::RenderPassColorAttachment {
+                    view,
+                    resolve_target: None,
+                    ops: wgpu::Operations {
+                        load: wgpu::LoadOp::Load,
+                        store: true,
+                    },
+                })],
+                depth_stencil_attachment: None,
+            });
+        }
+        
+        queue.submit([encoder.finish(), encoder2.finish()]);
 
         // If an error occurs, report it and panic.
         spawner.spawn_local(ErrorFuture {

Run the example (with RUST_BACKTRACE=1)

thread 'main' panicked at 'assertion failed: index < self.owned.len()', <removed>/wgpu/wgpu-core/src/track/mod.rs:355:9
stack backtrace:
   0: rust_begin_unwind
             at /rustc/f2d93935ffba3ab9d7ccb5300771a2d29b4c8bf3/library/std/src/panicking.rs:584:5
   1: core::panicking::panic_fmt
             at /rustc/f2d93935ffba3ab9d7ccb5300771a2d29b4c8bf3/library/core/src/panicking.rs:142:14
   2: core::panicking::panic
             at /rustc/f2d93935ffba3ab9d7ccb5300771a2d29b4c8bf3/library/core/src/panicking.rs:48:5
   3: wgpu_core::track::ResourceMetadata<A>::debug_assert_in_bounds
             at <removed>/wgpu/wgpu-core/src/track/mod.rs:355:9
   4: wgpu_core::track::texture::TextureUsageScope<A>::debug_assert_in_bounds
             at <removed>/wgpu/wgpu-core/src/track/texture.rs:237:9
   5: wgpu_core::track::texture::TextureUsageScope<A>::merge_single
             at <removed>/wgpu/wgpu-core/src/track/texture.rs:360:9
   6: wgpu_core::device::queue::<impl wgpu_core::hub::Global<G>>::queue_submit
             at <removed>/wgpu/wgpu-core/src/device/queue.rs:850:37
   7: <wgpu::backend::direct::Context as wgpu::Context>::queue_submit
             at ./src/backend/direct.rs:2291:15
   8: wgpu::Queue::submit
             at ./src/lib.rs:3497:19
   9: <cube::Example as cube::framework::Example>::render
             at ./examples/cube/main.rs:412:9
  10: cube::framework::start::{{closure}}
             at ./examples/cube/../framework.rs:375:17

(end of the backtrace removed because it's not particularly relevant)

Extra materials
Probably due to this line replacing used_surface_textures without ensuring it has enough capacity?

wgpu/wgpu-core/src/device/queue.rs

Line 964 in 72a4ef9

used_surface_textures = track::TextureUsageScope::new();

Platform
Linux with wayland (sway)
AMD Radeon RX 6700 XT (Vulkan)

[cwfitzgerald]

Oooo that's a scary one, thank god for debug asserts - will patch.