arm: Auto-vectorization for MVE: vabs

arm: Auto-vectorization for MVE: vabs

This patch adds support for auto-vectorization of absolute value
computation using vabs.

We use a similar pattern to what is used in neon.md and extend the
existing neg<mode>2 expander to match both 'neg' and 'abs'.  This
implies renaming the existing abs<mode>2 define_insn in neon.md to
avoid a clash with the new expander with the same name.

2021-06-03  Christophe Lyon  <christophe.lyon@linaro.org>

	gcc/
	* config/arm/mve.md (mve_vabsq_f<mode>): Use 'abs' instead of unspec.
	(mve_vabsq_s<mode>): Likewise.
	* config/arm/neon.md (abs<mode>2): Rename to neon_abs<mode>2.
	* config/arm/unspecs.md (VABSQ_F, VABSQ_S): Delete.
	* config/arm/vec-common.md (neg<mode>2): Rename to
	<absneg_str><mode>2.

	gcc/testsuite/
	* gcc.target/arm/simd/mve-vabs.c: New test.