Allow for reserved ranges to be plopped. Ie, selectively
call Reserve when plopping bytes instead of always doing it.

Prior to this, we placed dollops no matter now. Now,
we place dollops that are linked from pinned addresses.
This is very similar to the way that we placed
dollops in previous rewriter.

This fixes a bug where dollop creation incorrectly
continued past IBTs. IBTs should always start a
dollop and this fixes that.

Also, fix a small initializer bug in Dollop_t constructor

Move functionality to determine if a dollop for
an instruction already exists and whether to split
the containing dollop into AddNewDollop and use
that to implement UpdateTargets.

Any patch whose target immediately follows can be
turned into a five byte nop. This is not 100% ideal
but it is a good start to test whether this optimization
is worthwhile.