Fix hotkey suppression crash

[Filoppi]

EnableSuppression() was called in UpdateReferences() but then not executed if m_suppressor is already valid, causing input to stay nullptr in the m_suppressions map, which would fail to find an element and crash because it wasn't checked for validity.

I fixed the source of that happening, but also added safety checks around against nullptr.

I had gotten this was crash twice in months, both times in Debug, on startup, without doing anything, while Dolphin was loading (from the hotkey thread). It was probably just an order/timing dependency between threads.

If I had to guess, I'd say that it worked before because EnableSuppression() is called constantly by GetValue() so it would constantly refresh, but if things didn't happen in a specific order, it would crash.

The crash (on the it iterator) (callstack in on my branch, slightly different, but this bug was present and triggerable on master as well):

I've also included a small consistency fix for final_input_state being checked < and then > against CONDITION_THRESHOLD and not <= and then >.

Tested.

[iwubcode]

I'm confused why this force is needed. The only place you made it not the default is where the m_suppressor is already true?

[Filoppi]

EnableSuppression is called in two places and I didn't want to change the other call to it, so to keep changes to a min I added a force param which defaults to false.
Before it was calling EnableSuppression only if m_suppressor was true, but then the function inside would only proceed if it was false, which made no sense.