LLI Enhancement

This is just a note to mention that LLI will now correctly simulate the
endianness of the target machine (it's configured for sparc right now,
which is big endian), despite the fact that the machine you're running
lli on may be of different endianness (for example little endian on x86).

This fixes the last problem that I know of where running lli on x86
produces results that are different than lli on sparc. If you know of any
other problems, please let me know!