[llvm] Delete the incorrect assert that assumes size of `APFloat` is the same as `IEEEFloat`. (PR #111780)

[via llvm-commits]

Ariel-Burton wrote:

> This assertion is to make sure that EBO is performed on APFloat. What is the compiler you use to build llvm?

The issue is that APFloat has a union (`Storage`).  Downstream this includes a member / field for another floating point type.  We are in the process of upstreaming this type.  The representation of this type is larger than IEEE.  Note also that you are lucky that the `DoubleAPFloat` is not larger than `IEEEFloat`.

It's not a matter of the compiler used to build, but the assumption that no member of the `Storage` union is larger than `IEEEFloat` not being true.


https://github.com/llvm/llvm-project/pull/111780