R5900 Platform does not gracefully handle float literals being assigned to float variables

Version and Platform (required):

• Binary Ninja Version: 5.1.7821-dev
• Edition: Non-Commercial / Commercial / Ultimate
• OS: Windows 10 Pro
• OS Version: 10.0.19045 Build 19045
• CPU Architecture: x86-64

Bug Description:
When a float literal (such as 0x3F800000) is assigned to a float variable, instead of emitting HLIL such as var = 1.0f, HLIL such as __builtin_memcpy(dest: &var, src: "\x00\x00\x80\x3F", count: 4) is emitted instead.

Steps To Reproduce:
Please provide all steps required to reproduce the behavior:

1. Open an R5900 binary.
2. Change the type of a variable to float (assuming the variable is actually a float).
3. Observe the HLIL emitted by code assigning a value to that variable.

I've included a binary which demonstrates this issue (see @0x15be70)

Expected Behavior:
As per above, the HLIL should indicate that said float variable is being assigned that float constant.

Binary:
I've uploaded an archive to Portal with the key phrase: fierce forest sings calmly

This includes the original source binary and the bndb file.

Open the supplied R5900 binary and inspect HLIL at @0x15be70 after changing the variable type to float. Use the Portal binary and bndb with the reported reproduction steps, and confirm that the assignment is represented as a float literal rather than the emitted __builtin_memcpy form.