Vector loads/stores from/to primitive arrays and byte buffers involve bounds checks.

C2 already does some optimizations, but it's still not as good for scalar array accesses.

Investigate further into improving bounds check elimination for vectors to match performance of other array accesses.

(See the system property jdk.incubator.vector.VECTOR_ACCESS_OOB_CHECK defined in jdk/incubator/vector/VectorIntrinsics.java to control the generation of bounds checks.)