Add a test for orthogonal row-spanning table cells with collapsed borders.

The reference uses the same collapsed-border table with non-orthogonal content,
so it doesn't depend on how the collapsed borders are split between the cell and
the table.

Differential Revision: https://phabricator.services.mozilla.com/D342214

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=2060385
gecko-commit: 88b4dd4d70df75a55bf33cec240c75f7e7a5430c
gecko-commit-git: 6f13629728c5216ac1df9508b3686fb692c94ea0
gecko-reviewers: jfkthame

Add a csp-next web-platform-tests variant

Add a "csp-next" wpt variant which runs the content-security-policy tests with
the prefs of upcoming CSP features enabled, currently
security.csp.url-hashes.enabled and security.csp.strict-dynamic-url.enabled.
Each pref is exposed as its own run info property ("csp_url_hashes" and
"csp_strict_dynamic_url"), also used for updating metadata, so that test
expectations can be conditioned on the features they depend on.

The whole content-security-policy directory is tagged, so the tests run both
with the prefs disabled, in the default jobs, and enabled, in the variant. This
way, changes to the existing CSP behavior with the prefs disabled show up
separately from the new features. Once the prefs are enabled by default, the
variant can be changed to run with them disabled instead, until they're
removed.

The variant runs as tier 2, on Linux only, as the CSP behavior tested doesn't
depend on the platform.

Differential Revision: https://phabricator.services.mozilla.com/D337043

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=2086032
gecko-commit: e9f3fdf30480a989fb8efb0e3b258801a278bd2d
gecko-commit-git: 19e526c6488d50aa45f6949a296851b18049e5d6
gecko-reviewers: jgraham, taskgraph-reviewers, ahal

Use the SVG bounding box when scrolling an SVG element into view.

SVG container frames like don't set mRect, so scrolling one into view
(including focus()) used an empty rect at the origin of the outer . If
that point wasn't visible we would scroll even though the element itself was
fully visible. Use the same bounds getBoundingClientRect uses.

Differential Revision: https://phabricator.services.mozilla.com/D342865

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=2091995
gecko-commit: 1fe0d05053de3cbc58da29e53bf8fa726d66051c
gecko-commit-git: 1683369cd3504362320ae7ad658655c894312b79
gecko-reviewers: layout-reviewers, emilio

Add experimental RTCRtpEncodingParameters.encodingComplexity support

Implement the RTCRtpEncodingParameters.encodingComplexity WebRTC
extension behind the RTCEncodingComplexity experimental runtime feature
flag, mapping RTCEncodingComplexity enum values to
webrtc::VideoCodecComplexity.

This maps to w3c/webrtc-extensions#191

Bug: webrtc:42223911
Change-Id: I69653a67846eaa41f87fb59bc65ac66b9a992714
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/8759091
Commit-Queue: Erik Språng [email protected]
Reviewed-by: Guido Urdaneta [email protected]
Auto-Submit: Erik Språng [email protected]
Reviewed-by: Koji Ishii [email protected]
Cr-Commit-Position: refs/heads/main@{#1745014}

[webdriver][bidi] Test script.evaluate serialization of windows, platform objects, values from other realms and deep nesting

• window_proxy.py: same- and cross-origin iframe windows as results, in arrays,
  sets and maps, and thrown, and the window of a popup.
• platform_objects.py: platform objects without a more specific type serialize
  as {type: "object"} without a value, and a DOMException as an error. An object
  created with a platform prototype is an ordinary object, an observable array is
  an array, and page classes named Node, Element or NodeList don't change how
  nodes serialize.
• other_realm.py: arrays, maps, sets, dates, regular expressions, errors,
  promises, platform objects and nodes from an iframe's realm, and a shadow root
  that first gets a JavaScript object when it's serialized.
• nesting_depth.py, for script.evaluate and script.callFunction: deeply nested
  arrays and objects. A remote end can fail to serialize them, but it has to
  answer the command.
• exception_serialization_tentative.py: when the thrown value can't be
  serialized, the result has its type without contents. The spec doesn't say
  what to do in this case.

WebKit-Bug: https://bugs.webkit.org/show_bug.cgi?id=331201
WebKit-Canonical-Link: https://commits.webkit.org/326723@main

[webdriver][bidi] Test events for a subscription made with a child frame's context

session.subscribe resolves a child browsing context to its top-level traversable,
so the subscription covers the events of every frame in the tab.

• session.subscribe: browsingContext.load for a sibling frame, same and cross
  origin, for the top-level context after a navigation removes the frame, and for
  a nested frame.
• script.realmCreated: the existing realms of the tab's frames, and a new realm in
  a sibling frame. Also the realm of the document that a cross-origin navigation
  loads.
• script.realmDestroyed: the realm of a sibling frame.

WebKit-Bug: https://bugs.webkit.org/show_bug.cgi?id=330615
WebKit-Canonical-Link: https://commits.webkit.org/326419@main

Position out-of-flow boxes relative to their inline box containing block in both axes

An absolutely positioned box whose containing block is an inline box (e.g. ) is
positioned relative to the inline box in the axes with insets, but relative to the block that establishes the inline
formatting context in the axes using the static position. Everything that maps such a box has to pick the right origin
per axis.

Make the location relative to the border box of the inline box's first fragment in both axes, the same way it is
relative to the border box of any other containing block:

• the static position is converted into the coordinate space of the inline box, honoring its borders and right-to-left
  inline direction (staticDistance()),
• in vertical-rl and horizontal-bt, the conversion to physical coordinates happens within that border box (the new
  RenderBoxModelObject::blockStartRelativeToPhysicalForChild(), which RenderBox overrides),
• mapping adds the position of that border box in both axes (the new
  RenderBoxModelObject::blockStartRelativeFirstFragmentBorderBoxRect()), replacing
  PositionedLayoutConstraints::containingBlockOffsetForNonStaticAxes(), which only added it in the axes with insets,
• so do the places that add up locations on their own (webkit.org/b/330611).

This fixes, for boxes in an inline box containing block:

• the static position when the inline box has borders,
• right-to-left static positions,
• positions in vertical-rl,
• fixed positioned boxes in filtered inline boxes,
• the contribution to the scrollable overflow of the block.

WebKit-Bug: https://bugs.webkit.org/show_bug.cgi?id=330223
WebKit-Canonical-Link: https://commits.webkit.org/326189@main

[white-space-trim] Implement discard-inner on block containers

https://drafts.csswg.org/css-text-4/#valdef-white-space-trim-discard-inner

On block containers, discard-inner discards the white space at the beginning of the element up to and
including the last segment break before the first non-white-space character, and the white space at the end
starting with the first segment break after the last non-white-space character. This is what makes
"pre { white-space: pre; white-space-trim: discard-inner; }" drop the source formatting newlines around
the content while keeping its indentation. Only inline boxes were handled.

The fix trims the white space at the edges of the inline formatting context root. The rest of the change
is needed because the trimmed list can now be empty or have fewer items than the root has children:

• A block container with nothing but white space in it is left with an empty text item rather than an empty
  item list, which the inline formatting context takes for a list that has not been built yet.
• The simplified max-content path assumed that a single inline item meant a single child.

Block containers with block-level children are not covered yet: their inline content is in anonymous block
containers, which don't have the property.

WebKit-Bug: https://bugs.webkit.org/show_bug.cgi?id=331444
WebKit-Canonical-Link: https://commits.webkit.org/326772@main

[white-space-trim] discard-inner on an inline box leaves preserved white space in place

https://drafts.csswg.org/css-text-4/#valdef-white-space-trim-discard-inner

On inline boxes, discard-inner discards all white space at the beginning and end of the element, preserved
or not, while discard-before and discard-after only discard non-preserved white space. We only discarded
non-preserved white space, so the spaces, tabs and segment breaks that white-space: pre preserves stayed
("start[ some text ]end" under white-space: pre kept its spaces).

WebKit-Bug: https://bugs.webkit.org/show_bug.cgi?id=331443
WebKit-Canonical-Link: https://commits.webkit.org/326770@main

[white-space-trim] discard-inner trims the white space around an element that contains only white space

https://drafts.csswg.org/css-text-4/#valdef-white-space-trim-discard-inner

discard-inner discards the white space at the beginning and end of the element, but on an inline box
with only white space in it the trimming ran past the box's edges and took the white space before and
after the box too ("a b" rendered as "ab").

The fix replaces the single forward pass, which carried discard-inner's state past the box's end, with a
forward and a backward pass that each stop at the box's far edge.

WebKit-Bug: https://bugs.webkit.org/show_bug.cgi?id=331442
WebKit-Canonical-Link: https://commits.webkit.org/326766@main