Oftentimes a team at Google doesn't have the QA resources to test configuration X. In that case to avoid a potentially bad experience, feature Y is often gated to tested configs and and disabled for X, leaving X with the old experience since it had been tested on X.
Why does QA test only Chrome? Well if there is only enough QA time to test one browser, imagine having to be the guy to explain to the higher ups why it wasn't chrome.