Previously, the type names in the nonsemantic shader debug info would be "int", "uint", or "float" for all numeric types. This change makes the correct names such as "int8_t" or "float16_t" get emitted.
