This remove a theoretical performance optimization opportunity in that the UTF16 code path can elide code to check any trailing byte. This is leveraged in the aarch64 intrinsic. Performance evaluation on x86 and aarch64 shows that consolidating to only use the more generic StringLatin1::equals intrinsic is net neutral or even a net win.
This RFE suggest cleaning out the StringUTF16::equals code and associated intrinsic support.