This patch adds half-precision FP arithmetic and conversion instructions extending support of full-precision variants in v8.0.
FSQRT <Hd>.<Ts>, <Hn>.<Ts>
FSQRT <Hd>, <Hn>
SCVTF <Hd>.<Ts>, <Hn>.<Ts>
SCVTF <Hd>, <Wn>
SCVTF <Hd>, <Xn>
SCVTF <Hd>, <Hn>
SCVTF <Hd>, <Wn>, #<imm>
SCVTF <Hd>, <Xn>, #<imm>
UCVTF <Hd>.<Ts>, <Hn>.<Ts>
UCVTF <Hd>, <Wn>
UCVTF <Hd>, <Xn>
UCVTF <Hd>, <Wn>, #<imm>
UCVTF <Hd>, <Xn>, #<imm>
UCVTF <Hd>, <Hn>
Also adds bit manipulation instructions new in v8.2:
RAX1 <Dd>.2D, <Dn>.2D, <Dm>.2D
XAR <Dd>.2D, <Dn>.2D, <Dm>.2D, #<imm>
Issue: #2626