Currently we create a single nsDisplaySVGText item that just calls nsSVGTextFrame2::PaintSVG. We could instead have nsSVGTextFrame2::BuildDisplayList iterate over the TextRenderedRuns and generate a display item for each one. That should result in smaller invalidation areas when for example changing a single tspan's style.

For example if you load this and turn paint flashing on, you can see the whole rectangle that covers both tspans is invalidated, when layout.svg.css-frames.enabled = true. With the old text code, at least the two individual rectangles are invalidated. But we should be able to do better than both of these.