In moving a DCHECK around in https://chromium-review.googlesource.com/c/chromium/src/+/1079608 I discovered that some browser tests violate the assumption that the page scale layer includes only a scale transform.

interactive_ui_tests
All/AutofillInteractiveTest.FillThenFillSomeWithAnotherProfileThenClear/0

browser_tests:
TaskManagerViewTest.SelectionConsistency
PaymentRequestCreditCardEditorTest.EnteringValidDataInIncognito
PaymentRequestCvcUnmaskViewControllerTest.OpenGoBackOpenPay
PaymentRequestContactInfoEditorTest.ModifyExistingSelectsIt
PaymentRequestShippingAddressEditorTest.FailToLoadRegionData
PaymentRequestShippingAddressEditorTest.AsyncData
PaymentRequestCreditCardEditorTest.CreateNewBillingAddress
PaymentRequestIframeTest.IframeNavigation_Completed
PaymentRequestIframeTest.HistoryPushState_Completed
PaymentRequestNoUpdateWithTest.BuyWithoutPromises
PaymentRequestShippingAddressEditorTest.SyncData
PaymentRequestContactInfoEditorTest.ModifyExisting
PaymentRequestCreditCardEditorTest.DoneButtonDisabled
PaymentRequestJourneyLoggerMultipleShowTest.ShowSameRequest
PaymentRequestContactAddressUseStatsTest.RecordUse

trchen@ had this to say:
"""
Hmmm, this is the result from a weird hack:
https://cs.chromium.org/chromium/src/cc/trees/layer_tree_host.cc?rcl=16d6164b1b7a2bc7a6920be8af1b59b947fe3c99&l=742

Where if a client doesn't specify page scale layer, the parent of the first scrollable area will be used. It is only used in property tree builder, and even LayerTreeImpl doesn't know it. The whole thing seems very wrong.

Have you tried if anything breaks with that hack removed?
"""