Mark webgpu:api,operation,buffers,map_oom:map* as failing

These were marked as failing on only Windows and Linux. A Dawn change
is coming which will update the Metal backend to also throw OOM on large
buffer allocations. Pre-emptively marking as failing to prevent test
failures. The main issue here is that the test incorrectly does not
allow the implementation to throw an OOM error on enormous buffer
allocations.

Bug: dawn:433
Change-Id: Id5123ea1608bf68916a5c40d2d682f8c715e3a0f
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2225792
Commit-Queue: Kai Ninomiya <kainino@chromium.org>
Reviewed-by: Kai Ninomiya <kainino@chromium.org>
Cr-Commit-Position: refs/heads/master@{#774014}

third_party/blink/web_tests/WebGPUExpectations

@@ -52,6 +52,11 @@ crbug.com/dawn/375 wpt_internal/webgpu/cts.html?q=webgpu:api,validation,createBi
 crbug.com/dawn/375 wpt_internal/webgpu/cts.html?q=webgpu:api,validation,createBindGroup:buffer_offset_and_size_for_bind_groups_match:offset=1024;size=0 [ Failure ]
 crbug.com/dawn/375 wpt_internal/webgpu/cts.html?q=webgpu:api,validation,createBindGroup:buffer_offset_and_size_for_bind_groups_match:offset=1024;size=undefined [ Failure ]
 
+# Test doesn't check for an OOM error. gpuweb/cts#199 crbug.com/1014740 crbug.com/1014740
+# Actually passes on Windows NVIDIA which may be a Dawn bug.
+wpt_internal/webgpu/cts.html?q=webgpu:api,operation,buffers,map_oom:mapReadAsync: [ Failure Crash ]
+wpt_internal/webgpu/cts.html?q=webgpu:api,operation,buffers,map_oom:mapWriteAsync: [ Failure Crash ]
+
 # Timeout on Windows/NVIDIA with backend validation
 wpt_internal/webgpu/cts.html?q=webgpu:api,validation,createBindGroupLayout:max_resources_per_stage,in_pipeline_layout:* [ Failure Timeout ]
 
@@ -78,9 +83,6 @@ crbug.com/1083478 [ Mac ] wpt_internal/webgpu/webgpu/web-platform/reftests/canva
 # Linux (Vulkan) specific
 #
 
-crbug.com/1014740 [ Linux ] wpt_internal/webgpu/cts.html?q=webgpu:api,operation,buffers,map_oom:mapReadAsync: [ Failure Crash ]
-crbug.com/1014740 [ Linux ] wpt_internal/webgpu/cts.html?q=webgpu:api,operation,buffers,map_oom:mapWriteAsync: [ Failure Crash ]
-
 crbug.com/dawn/399 [ Linux ] wpt_internal/webgpu/cts.html?q=webgpu:api,validation,createBindGroup:texture_binding_must_have_correct_usage:type="sampled-texture";* [ Crash ]
 crbug.com/dawn/399 [ Linux ] wpt_internal/webgpu/cts.html?q=webgpu:api,validation,render_pass_descriptor:* [ Crash ]
 
@@ -101,7 +103,3 @@ crbug.com/dawn/323 [ Win ] wpt_internal/webgpu/cts.html?q=webgpu:api,validation,
 
 # Very flaky on Windows/Linux, especially (but not exclusively!) with backend validation
 crbug.com/1087130 [ Win ] wpt_internal/webgpu/cts.html?q=webgpu:api,validation,createView:* [ RetryOnFailure ]
-
-# These fail on NVIDIA (but it's probably our bug)
-crbug.com/1014738 [ Win ] wpt_internal/webgpu/cts.html?q=webgpu:api,operation,buffers,map_oom:mapReadAsync: [ Failure ]
-crbug.com/1014738 [ Win ] wpt_internal/webgpu/cts.html?q=webgpu:api,operation,buffers,map_oom:mapWriteAsync: [ Failure ]