[SR-13899] [Sema] Adjustments on coerce to checked cast diagnostics

[LucianoPAlmeida]

Based on https://forums.swift.org/t/missing-conditional-cast-fix-it/42282
This proposes to adjust coerce to checked cast diagnostic to be broken into the diagnostic and fix-it note.
Also, cases were the coercion result has a value to optional restriction e.g. conditional binding

let any: Any = 1
if let int = any as Int { // an `as!` here would lead to another malformed code because conditional binding expects an optional type

}

we suggest conditional downcast as? instead of force downcast.

The break into a note is more like a proposal not really the SR, so let me know what you think :)

cc @dan-zheng

Resolves SR-13899.

[LucianoPAlmeida]

@swift-ci Please smoke test

[dan-zheng]

Thanks for the holiday gift, you are our secret 🎅

[xedin]

Looks good but it would be great if we could detect that conditional cast could be used during solving, would it be possible? That way we could use a special fix/diagnostic for it instead of adding that for "forced" downcast which is kind of misleading...

[LucianoPAlmeida]

That way we could use a special fix/diagnostic for it instead of adding that for "forced" downcast which is kind of misleading...

Humm, do you think is worth another fix/diagnostic for that? Maybe we could make the "forced" downcast more generic and if we detect that conditional cast could be used when solving pass as an additional parameter to the fix and diagnostics. WDYT?

[LucianoPAlmeida]

@xedin Let me know what you think of this approach :)

[xedin]

Looks much better! I have left a nit comment about naming of the failure, feel free to merge once that's fixed.

[LucianoPAlmeida]

@swift-ci Please smoke test

[LucianoPAlmeida]

Thanks @xedin!