follow-up: To followup on our discussions around cx and the middleware asserting it...
I really doubt the usefulness and even challenge the correctness of passing cx for such action!

We do pass cx for this action:
https://searchfox.org/mozilla-central/rev/b579290e6b830d1b3f0a941203b0c0e1e56c42a3/devtools/client/debugger/src/actions/file-search.js#61-63

The main reason to pass cx in actions is to have this middleware to stop the action if the context changed:
https://searchfox.org/mozilla-central/rev/b579290e6b830d1b3f0a941203b0c0e1e56c42a3/devtools/client/debugger/src/actions/utils/middleware/context.js#19,27
And this will typically throws and prevent the action from reaching the reducers if we navigated:
https://searchfox.org/mozilla-central/rev/b579290e6b830d1b3f0a941203b0c0e1e56c42a3/devtools/client/debugger/src/utils/context.js#52-58
https://searchfox.org/mozilla-central/rev/b579290e6b830d1b3f0a941203b0c0e1e56c42a3/devtools/client/debugger/src/reducers/pause.js#308-313

But here... imagine for some weird reason we do navigate while toggling modifiers.
If the use clicked on the modifiers, we do want to unconditionaly toggle the state in the reducer.

It is really unclear this particular action have been affected by https://bugzilla.mozilla.org/show_bug.cgi?id=1535362 ...

Also, the middleware assertion is especially useful when things are asynchronous.
Here the action is synchronous, it isn't even a thunk action.
And we might assume that React components are somewhat synchronously correct? But I could imagine some ways where React is somewhat async and it may, as async thunk action, benefit from the middleware assertion...