better encoding failure error messages

_From [bruen...@google.com](https://code.google.com/u/109494838902877177630/) on November 20, 2011 09:55:30_

this was PR 250947

Even for me, it can be difficult to figure out why a particular high-level
instr_t does not encode. I end up spending several minutes in the
debugger.

There are some encoding errors that are very confusing if you can't see our decode_table
templates:
- xchg reg,mem: memory MUST be the destination, not the source
- subtle size differences in operands: the biggest problem. PR 225845 tried to make
this easier for confusingly-sized operands.

It's not easy to give back good errors w/o walking back through the
encoding process and reporting on every template match attempt, but we
should consider doing that.

_Original issue: http://code.google.com/p/dynamorio/issues/detail?id=614_