[BUG] Incorrect gating of FADD, FSUB (accumulate into ZA)

[rockdreamer]

Describe the bug

In the "ADD, SUB, BFADD, BFSUB, FADD, FSUB (accumulate into ZA)" section, we guard half precision element operations using __ARM_FEATURE_SME_F16F16 only.

However, "Two ZA single-vectors of half-precision elements" and "Four ZA single-vectors of half-precision elements" encodings are available when either (FEAT_SME_F16F16 || FEAT_SME_F8F16) features are available.