I've seen the PR: #6589 and I got excited for this new feature because I don't need to calculate legend positioning for each view size.
But on this line here: 5c94e46#diff-5da3ec3a0416c8572b65cf31fe31dbc248f8791065a6749405076431efea1179R173 it says dflt: 'container', with I guess is default value: "container" ?
But on this pen: https://codepen.io/hannahker/pen/wvEZZvN , if you comment "xref": "container", or "yref": "container", or replace the values with paper you can see that the default ref is container.
Is this intended or not ?