The assumption from the start has been that we would only ever want to show or see oneof
Nope. Not at all. It is not at all about wanting to see only one pane. Heck, any design which allows undocking as Manifold does is clearly not prejudiced against letting people have many windows floating about to see all sorts of things at once. Manifold loves the idea of people being able to add displays to the flight deck of their GIS starship.
Instead, the issue is entirely about the response to user actions a non-modal pane must have. That is difficult enough with just one non-modal pane, so difficult that the whole reason people reflexively default to modal dialogs is to dodge those difficulties. When a modal dialog is open, it seizes the user interface. No need to wonder which window responds to whatever the user does and to sort out possible conflicts, because you know it is just the one modal pane that has seized all focus.
Non-modal panes are very different: you can build interfaces that are fluid and natural with one non-modal pane, but the user interface rules become so intricate with more than one pane that, historically, the cost of educating users in the labyrinth of limitations and special case behavior rules that arise is far greater than what you gain by having more than one pain.
For example, it would often be useful to keep Component and/or Layers visible, while we are editing Styles. Or to keep Record visible, when adjusting Select. Or to keep Select visible, when performing a Transform.
I agree that would be useful, but as reporting widgets/displays, not as active panes. Perhaps an alt-click on the panel's caption could undock it into reporting gadget. That is a new interface idea, to allow panes to be undocked as viewbots of sort, which report their contents but which are not active as panes if they are not clicked into activity (at which point the previous pane automatically switches to read-only display widget status).
There are very many nuances to that idea that take a lot of time to think about and engineer, as the idea of a display that sometimes is a non-modal pane and sometimes not is a very non-routine thing. I'm not even sure it is possible within the sort of engineering expense that makes sense to apply to such a thing (for example, if the choice is between an Internet Map Server now or this thing as being the same engineering effort, we would do IMS).
For now, we don't do that because it is one non-modal pane that is the listener to user actions, switched into whatever regime is the right one to respond. The automatic switching by context (for example, the Record panel popping to the fore when you alt-click an object in a drawing) works naturally within that system as well. It works very well despite the complexity of context switching in a parallel system, etc. As you've pointed out, there are bigger fish to fry.
As for the more compact design for Contents pane panels, well, I like it a lot and it is a simple matter to implement. A big reason for it can be seen in the following two illustrations, the old and the new.
The old design wastes an enormous amount of screen real estate for blank captions. That's a crime for small screens and dead space/distraction even on large screens. I grant a single click on the panel you want is quick, but a click-click to the panel you want is pretty darned near as quick given the proximity of the clicks.
The ctrl-<number> shortcuts are a plus for those who want them, not a key part of the UI. Like all second or third tier keyboard shortcuts they aren't for everyone, but having them is no downside whatsoever for those who do not use them. If you don't use them, the new interface is plenty fast.
I understand that for beginners or people who haven't used the system for a while there is a certain utility in gadgeting up the interface with all the possible choices always on display. But it takes just a click to remind yourself what is there (third screen below), if you use the system for more than a few minutes you'll be refreshed, and having all the options hanging around as blank captions is both expensive in real estate and also distracting, not at all "quiet cockpit" for regular users.
How this thread started is a good illustration of the expense and distraction: no point in being wasteful of real estate because some of us use three or more monitors. A compact, efficient interface is also good for many monitors, and it becomes a life-saver with single screens and especially the smaller devices often used in the field.
The old interface on left and new interface on right:
The new interface with a single-click anywhere on the caption bar...
I remind everyone that in the old design you had to click at least once on a caption to switch from one panel to another, and often the click could be far apart, at the bottom of a pane or the top. In the new design you have to click twice, but the clicks are close together all at the top.