Add little-endian octeon mips arch

[jonathanzetier]

I didn't think these actually existed in the wild, but lo and behold, we have stumbled upon a little-endian cavium mips64 ELF. Unfortunately I can't share it, but I can confirm that this properly disassembles the binary we discovered, whereas before, it would be appear as total garbage because it was trying to decode little-endian MIPS as big-endian.

I'm pretty sure I added the little-endian variant everywhere that had the big-endian variant, but please double-check and let me know if there's somewhere I missed!

[fuzyll]

I've been working on a thing in the background for awhile that lets me generate binaries of different architecture/platform combos and I can confirm that this makes a sample binary that does return 0; go from:

004004b0    int64_t main(int64_t, int64_t, int64_t, int64_t arg4, int32_t arg5 @ $t7)

int64_t arg4  {Register $a3}
int32_t arg5  {Register $t7}

004004b0     0  f0ffbd67   scd     $ra, -0x4299($a3)  {0x1}  {0x0}
❓004004b4     0  0800beff   j       0x2fbfc
004004b8     0  25f0a003   addiu   $s0, $t7, -0x5ffd

...to:

004004b0    int64_t main() __pure

void var_10  {Frame offset -10}
int64_t __saved_$fp  {Frame offset -8}
void arg_0  {Frame offset 0}

004004b0     0* f0ffbd67   daddiu  $sp, $sp, -0x10
004004b4    10  0800beff   sd      $fp, 8($sp) {__saved_$fp}
004004b8    10  25f0a003   move    $fp, $sp {var_10}
004004bc    10  25100000   move    $v0, $zero  {0x0}
004004c0    10  25e8c003   move    $sp, $fp
004004c4    10  0800bedf   ld      $fp, 8($sp) {__saved_$fp}
004004c8    10* 1000bd67   daddiu  $sp, $sp, 0x10
004004cc     0  0800e003   jr      $ra
004004d0     0  00000000   nop     

Which, as far as I can tell, means we've gone from "not real code" to "real code". Seems like an obvious win to me.

Will get this merged as soon as I can. 😄