Partial in-lining handles copy and mismatch of array sizes less than 64 bytes through JIT code rather than calling optimized stubs thereby saving costly call overhead.
Enable partial in-lining optimization for AMD EPYC Genoa and Turin targets which support AVX512 features.

Currently -XX:ArrayOperationPartialInlineSize is set to zero for these targets.