Add lifting for vector comparisons, bitwise operations, widening and
narrowing moves, multiply-long, negation, pairwise addition, shifts,
swaps, transposes, tests, and interleaving operations. Use dedicated
intrinsics for lane-wise operations and explicit IL for bitwise
operations, register swaps, and scalar floating-point negation.

Correct SIMD decoding, signedness, immediate types, D/Q operand widths,
and reserved encodings. Support ARMv8 VMULL.P64 polynomial products.

Add FLDMDBX/FLDMIAX and FSTMDBX/FSTMIAX lifting, accounting for the
unused trailing word in addressing and writeback. Correct Thumb
decoding of these forms, including SP-based transfers.