x86 jump table detection can break in some cases

[cyanpencil]

Our x86 jump table detection works like this: for every array a accessed by the code through a rip-relative load, if &a is in .rodata, and &a + a[0] is the address of an instruction, then a is a jump table.

This needs some data flow analysis that checks that the value loaded from a is used into an indirect branch.

If possible, it would be cool to check that the code that accesses a has some upper bounds checking on the register used for a's index (which is basically a check for the 'default' case in jumptables)

An example of an array in memory that is falsely detected as a jump table is Case 1 of this issue (HexHive/retrowrite#20)

[diagprov]

Suggest we move as an enhancement on https://github.com/HexHive/retrowrite?