Update serialization_lib submodule

- using native size of bfloat/float8 serialization/deserialization
- updating cfloat.h
- fixing bfloat16 data conversion for serialization

Signed-off-by: Won Jeon <won.jeon@arm.com>
Change-Id: I8e8c10a3a9466f47d73d8a29be0a7d2467f51696