WASM only supports Little-endian byte ordering and we need a mechanism to
reverse the ordering efficiently on Big-endian machines.
Up until now this was done using TF graphs within wasm-compiler.
The new approach allows for having more machine level optimizations
by introducing the new "kSimd128ReverseBytes" opcode which gets executed
only on Big-endian machines.

Change-Id: I63c6c3c42ca9ff9d9b2af2d45070a70cf1b3cefc
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1803494Reviewed-by: Michael Starzinger <mstarzinger@chromium.org>
Commit-Queue: Milad Farazmand <miladfar@ca.ibm.com>
Cr-Commit-Position: refs/heads/master@{#63875}