Use full move for byte registers stores in jumps. (dotnet/coreclr#11570)
* Use the 4-byte move for jump spilling.
The other types of spilling already use the proper move:
1) for lsra spilling it always use 4-byte move because we allocate 4-byte slots;
2) for other types it sets needsByteReg and lsra chooses correct register;
We do not apply the second approach to fixing this issue because jmp doesn't
have real uses, that can keep this requirement on. Also, it creates more strict
restrictions, that we need.
Commit migrated from https://github.com/dotnet/coreclr/commit/
81baf0a53fdc3cae247dd80ebc1e8f54e7ad54fa