[clang] [CIR] Mark bit-field access units as their own member kinds (PR #216864)

[Erich Keane via cfe-commits]

https://github.com/erichkeane commented:

I THINK I understand data vs bitfield, but why empty vs bitfield-empty?  I thought the 'empty' was supposed to represent fields we couldn't access anyway? How are these different? 

https://github.com/llvm/llvm-project/pull/216864