So we can discount the keyboard entirely I think.
No, since a lot depends on what the keyboard is doing internally. Keyboards all now depend on internal programs that run the microcontroller deciding what key presses do, so what the keyboard decides to do when a Shift key and Esc key are pressed together is up to that in-keyboard programming.
A further issue is that the in-keyboard programming might not be interacting with the driver correctly in all cases. That's especially true of keyboards that come with so-called "value added" features.
I got curious about this and did some searches, and there's no end of hits for problems with the Esc key and Shift-Esc sequences with various HP laptops, and Logitech keyboards, as well as with other brands.
As a practical matter, since Microsoft calls the shots on what happens with Windows, the test is to see what Microsoft keyboards and mice do. If you take a clean, freshly loaded Windows system and run your application with a Microsoft keyboard and mouse, if it works fine that's a useful indication it is interacting with Windows correctly.
It's also an indicator that a keyboard is the problem if it works fine, stops working, and then re-installing the driver makes it work again.
I personally don't like Shift-Esc because I'm a touch-typist and that's an awkward combination. I'd prefer Manifold just used Esc. But in any event, the hardware in use shouldn't monkey around with doing anything but providing an honest, unmodified Esc and Shift-Esc if that's what the user presses.