Support for 3D ancestor transforms in getScreenCTM()

[BigBadaboom]

https://svgwg.org/svg2-draft/single-page.html#types-__svg__SVGGraphicsElement__getScreenCTM

Should the definition of getScreenCTM() make any comment on how 3D transforms on ancestor (HTML) elements should be handled? From my experimentation, it seems Chrome does attempt to support this (but is currently bugged). Firefox does not seem to have updated getScreenCTM() for SVG2 yet.

https://jsfiddle.net/r8887x05/1/

<div id="mydiv" style="transform: translate(1px,0) scale(2)">
  <svg>
    <circle id="mycircle" cx="50" cy="50" r="10"/>
  </svg>
</div>

var div = document.getElementById('mydiv');
var circle = document.getElementById('mycircle');
console.log("before=", circle.getScreenCTM());
div.setAttribute("style", "transform: translate3D(1px,0,0) scale(2)");
console.log("after=", circle.getScreenCTM());

Chrome results (bug? scale transform is reflected in e and f, but not a and d):

before= SVGMatrix { a: 1, b: 0, c: 0, d: 1, e: -335.5, f: -69 }
after= SVGMatrix { a: 1, b: 0, c: 0, d: 1, e: -335.5, f:-69 }

Firefox results (ancestor transforms have no effect):

before= SVGMatrix { a: 1, b: 0, c: 0, d: 1, e: 8, f: 8 }
after= SVGMatrix { a: 1, b: 0, c: 0, d: 1, e: 8, f: 8 }

[boggydigital]

Not blocking updated 2.0 CR publication - assigning 2.1 WD milestone

[dirkschulze]

We do have another issue with vector-effects where this question came up. During the testing, it seemed like browsers do cross the pseudo-3D boundaries. Needs testing.

[dirkschulze]

Related to #582