Having the UI elements (screen canvas) rendered mixed with world objects in Scene View is really impractical and counter-intuitive.
The 3D world objects and the 2D UI elements are shown in the same coordinate system, in a way that is not how they would actually appear in the game, with objects from both potentially obscuring each other. So you often have to switch 2D mode on and off and sometimes layers too, and also pan and zoom accordingly, in a impractical way.
My suggestion is to have two coordinate systems rendered in the Scene View: the "Screen Space" (always rendered in 2D mode) for screen canvases and the "World Space" (can be rotated in 3D normally) for world objects.
Add a toggle in the toolbar to switch the active space. All interaction, like zooming, panning, selecting-by-click, grid lines, etc would apply only to the active space. The screen space would show rendering for the display selected in a dropdown.
If possible, add another dropdown in the toolbar, or maybe some other way to select which canvases are rendered in the Screen Space of the Scene View. Canvases that are on "Screen Space - Overlay" would be toggled off/on. Canvases on "Screen Space - Camera" would be toggled between off, editable in Screen Space (in 2D mode) and in World Space (3D, as a plane in front of the camera, like they are currently).