Currently, we fail WPT tests aspect-ratio-intrinsic-size-001.html through -006, as shown here:
https://wpt.fyi/results/css/css-flexbox?label=master&label=experimental&aligned&view=subtest&q=aspect-ratio-intrinsic-size

Directly link to the first testcase:
https://wpt.live/css/css-flexbox/aspect-ratio-intrinsic-size-001.html

It looks like the general setup is:

• The flex container has a specified pixel-valued height, and a content-based width.
• It has one flex item which gets the default align-self:stretch behavior, so it's as tall as the flex container.
• That flex item has an aspect ratio (or has a child with an aspect ratio)
• The tests expect the flex item to use its stretched height when determining its intrinsic-width-contribution to the flex container's intrinsic width. i.e. it expects the flex container to end up being a square, with its specified height being transferred through to its width (via vertically stretching the flex item and thereby influencing its horizontal intrinsic size contribution)

(adding see-also since this is similar to bug 1816680, though bug 1816680 is focusing on the sizing of the flex item -- over there, the flex container does not have a content-based width.)