Two types of zoom lenses exist. The first is the parfocal type which is designed to keep focus throughout its zoom range (the type the OP has). Interestingly, these are "tuned in" to track focus by fine adjusting the back focus spacing between the lens, and the film or sensor (in the design and manufacturing process). This is what the OP did away with by adding an extension tube, and therefore the lens will no longer track focus as it's zoomed. As mentioned, adding a teleconverter
shouldn't affect a parfocal's ability to hold focus, but it can happen with flaws in the design of a teleconverter because the back focus spacing of the main zoom is so critical to focus tracking.
The second type of zoom is what is usually seen today in still camera zooms, and it is a vari-focus type which has to be refocused whenever the zoom is adjusted. The latter doesn't relate to the OP's issue but as a side note, it finds use today because of auto-focusing cameras which will adjust the focus each time they take a shot.
You still see parfocal zooms ($$$) used extensively in cinema production because they need to track as zooming is adjusted. Otherwise the focus puller would have an incredibly difficult job keeping them in focus throughout a zoom shot.