[Sema][SR-14408] Improvements on enum equality diagnostics

[LucianoPAlmeida]

For some cases when attempting an == overload choice for enum and a dot member argument, for some choices solver ends trying to record a missing member for the argument. This causes the end result to be a set of solutions where although the score is better, given the context are not exactly. This causes diagnose ambiguity with fixes to work with a set of "best score" solutions that would not give the best diagnostics.
So the idea here is to look at the context in which this fix is being recorded and increase the impact if in this context a missing member fix less likely to be an useful diagnostic on this case.

Resolves SR-14408.

[xedin]

I think it would make sense to increase an impact of missing member fix in general (at least for argument position), so it doesn't interfere in situations where there is more than one specific fix. What to give it a try?

[LucianoPAlmeida]

I think it would make sense to increase an impact of missing member fix in general (at least for argument position), so it doesn't interfere in situations where there is more than one specific fix. What to give it a try?

Sure! Let's try this :)

[LucianoPAlmeida]

I think it would make sense to increase an impact of missing member fix in general (at least for argument position), so it doesn't interfere in situations where there is more than one specific fix. What to give it a try?

Adjustments made @xedin
Run tests locally and I'm surprised that this change didn't affect any other diagnostic

[xedin]

Looks good!

[xedin]

I expected that not much is going to get effected there since missing member is de-prioritized from the get go.

[xedin]

[LucianoPAlmeida]

I expected that not much is going to get effected there since missing member is de-prioritized from the get go.

Ah true, makes sense. Thank you!

[LucianoPAlmeida]

@swift-ci Please smoke test