Reid suggested making HasWinCFI a plain bool defaulting to false in D50288.
It's needed in order to add HasWinCFI to MIRPrinter (D51201 adds HasWinCFI to MIRParser only). Otherwise, we'll get an assertion
HasWinCFI.hasValue() && "HasWinCFI not set yet!"'
Also, a few ARM64 Windows test cases will fail with the same assert if the ARM64 MCLayer part (D50166) goes in before the frame lowering part that sets HasWinCFI (D50288 as of now).