Possible incorrect result in compiler-rt 128x128 `wideMultiply`

[tgross35]

This bit of code seems to have a bug where it is missing the carry bit. This was discovered when I was porting the function to Rust's compiler-rt and results of unit tests disagreed with Python, see rust-lang/compiler-builtins#587 (comment) and the followup discussion.

To reproduce - "0xffffffffffffffffffffffffffffffff * 0xffffffffffffffffffffffffffffffff should be 0xfffffffffffffffffffffffffffffffe00000000000000000000000000000001, but this function returns 0xfffffffffffffffffffffffffffffffd00000000000000000000000000000001.

cc @marthadev who found a fix for the Rust side.

Appears to cause #91840 (comment)

[tgross35]

I suspect this just hasn't come up because a u128 is never saturated by the mantissa of a f128