Arm: Enable SVE intrinsics by default if compiler supports it

In #443 we added support for SVE and SVE2 feature detection. This was
initially disabled by default as there were no optimized implementations
using SVE at the time.

Now that we have quite a few SVE optimized functions, enable SVE by
default if the compiler supports it. SVE2 is left disabled by default
since there is still no code making use of it yet.

Benchmarking on a Neoverse V2 machine with LLVM 20, this improves
performance of a --preset=medium encoding by ~3.7% over the previous
default configuration.

Author: georges-arm

CMakeLists.txt

@@ -48,7 +48,7 @@ endif()
 # We enable x86 intrinsics for all target architectures, because they are implemented through SIMD-everywhere on non-x86.
 set( VVENC_ENABLE_X86_SIMD TRUE                      CACHE BOOL "Enable x86 intrinsics" )
 set( VVENC_ENABLE_ARM_SIMD ${VVENC_ARM_SIMD_DEFAULT} CACHE BOOL "Enable Arm Neon intrinsics" )
-set( VVENC_ENABLE_ARM_SIMD_SVE FALSE                 CACHE BOOL "Enable Arm SVE intrinsics" )
+set( VVENC_ENABLE_ARM_SIMD_SVE ${FLAG_sve}           CACHE BOOL "Enable Arm SVE intrinsics" )
 set( VVENC_ENABLE_ARM_SIMD_SVE2 FALSE                CACHE BOOL "Enable Arm SVE2 intrinsics" )
 set( VVENC_ENABLE_LOONGARCH64_SIMD_LSX ${VVENC_LOONGARCH64_SIMD_DEFAULT} CACHE BOOL "Enable LoongArch64 LSX intrinsics" )
 set( VVENC_FFP_CONTRACT_OFF OFF                      CACHE BOOL "Disable floating point expression contraction fFr exact math" )