And don't forget all the really nasty political things going on in the backstage of big 3 journals (Nature, Science & Cell). Big PIs often reach agreements with editors before writing their papers.
What to me would sound less unlikely (and I am totally on the outside of all of this), if PIs take into account editor's interests and preferences -- that is, if you know that editor is interested in topic A (say, oxygen on exoplanets) you might lean in that direction with the paper you hope to publish there (e.g., in abstract / intro / conclusion stress how your research supports existence or lack thereof of oxygen). Also not impossible to know if an issue is dedicated to a specific grand theme, so articles on that have a significantly higher acceptance chance.
Whether this is 100% kosher or not can be debated, but this is far, far off from agreements that a journal will publish whatever a specific researcher writes.
If you need anecdotal examples, I'm sure we can provide some :)
If you're a PI on a big project or heading a big group, it's routine business to meet journal editors, invite them to your lab and reach publication agreements before writing your paper.
Obviously the paper will still go to review. But even that is already a big hurdle you have bypassed. In the big 3 journals, it's equally tough to get to review than to pass reviews once there. Plus, with direct contact to the editor, it's easier to get your paper reviewed by friendlier peers and/or influence ultimate editorial decisions.