[Bug] DUMMY_MOD_NEUTRALIZER doesn't work as intended about qmk_firmware

[iQuerz]

@precondition this is the issue I've mentioned on the discord server. Thank you for looking into it :)

[precondition]

As far as I know, the code of the key overrides feature is not called when the modifier is released so I suspect that patching this issue will require turning on a bitset flag in quantum/process_keycode/process_key_override.c and checking for the activation of that flag in the core key processing code (e.g. in pseudo-code if (key.is_mod() && !key.pressed && MOD_BIT(key) & neutralized_mods) { tap_code(DUMMY_MOD_CODE_NEUTRALIZER); } unregister_code(key);)

Ideally, we would find a solution that only requires changes to the process_key_override.c file but I don't know if that's possible.

Another option would be to avoid re-enabling the modifier in the first place, which has the benefit of being easy to implement within process_key_override.c. However, this approach would break the following use case illustrated below:

Physical key	Physical key status	Keyboard report to the host
Left Alt	down	Left Alt
9	down	Left Alt+DUMMY_KC → Home
9	up	Left Alt
K	down	Left Alt+K
K	up	Left Alt
Left Alt	up	<empty>

In the hypothetical case where an Emacs user would want to move the cursor to start of the line with Alt+9 converted to Home firmware-side and then cut the entire sentence after the cursor with the Alt+K Emacs shortcut.

[sigprof]

One option here is to emit the neutralizer in clear_active_override() if we are reregistering an affected modifier which was suppressed.