Floating point type propagation failure

[plafosse]

MLIL fails to properly propagate a floating point type to the constant under some conditions.

In the included binary you can see that on instruction 0x0000298f the constant isn't being represented as a floating point value and the zmm0 variable doesn't have the type of double

It should be trivially determinable via the following instruction indicating its part of a floating point add

fp_bug.zip

Special Thanks to: Zao Yang and Stefan Nagy for their research in Decompiler Fuzzing for reporting this issue.