[Gecko Bug 2018114] Part 3: Add more tests for lspace/rspace and update expectations.

[moz-wptsync-bot]

• mo-lspace-rspace-5.html: This is a new test to actually verify the
  spacing around (embellished) operators that have lspace/rspace,
  using the distance to the previous/next sibling box for 8 cases
  (all combinations of math/mrow container, RTL/LTR directions and
  embellished/non-embellished operators). For each of them, we
  try various kinds of operarors (normal text, largeop, stretchy
  variant and stretchy glyph assembly). In the past we were always
  including the lspace/rspace in the width of the (embellished)
  operators, so the actual distance was zero and all the tests were
  failing. Now, we only do that for children of math/mtd.

• painting-stretchy-operator-002.html: This is a new test to verify
  that an operator with lspace/rspace is painted at the proper
  horizontal offset, for 8 cases (all combinations of math/mrow
  container, RTL/LTR directions and embellished/non-embellished
  operators). This is already passing in all browsers even when they
  wrongly include lspace/rspace in the operator's width, but it allows
  to catch potential painting regressions (especially the need for the
  new nsMathMLmoFrame::FixInterFrameSpacing).

• no-spacing.html: This test verifies that spacing is not added around
  <mo> elements with lspace/rspace, when the parent does not perform
  mrow layout ; and that spacing is added around the parent if and only
  if that parent is an embellished operator. In the past we were always
  modifying the width of (embellished) operators so the width-based
  checks of the test were failing. Now, we only do that for direct children
  of math/mtd, not for (non-embellished) mrow elements. We extend these
  tests to cover these three kinds of parents. Additionally, we were
  also always adding spacing around mroot (which is never an
  embellished operator), but this is fixed now.

• mrow-preferred-width.html: This test verifies preferred width of
  mrow, including the cases when it contains a stretchy operator with
  lspace/rspace. Although we currently pass it, the preferred width is
  currently wrong when the operator is actually embellished, so we
  extend the test to cover that case. Additionally, we add cases
  when the mrow-like container is actually the tag.

Differential Revision: https://phabricator.services.mozilla.com/D284372

bugzilla-url: https://bugzilla.mozilla.org/show_bug.cgi?id=2018114
gecko-commit: 043279b5701d8d3dc67c994e0d4ebbb334905206
gecko-reviewers: emilio

[wpt-pr-bot]

The review process for this patch is being conducted in the Firefox project.