Let a plugin tell us about where to plop a dollop.
This change includes a

1. utils namespace with a function to calculate
the worst case instruction size.
2. Dollop_t type that can be used to describe
and pass around a dollop.

Remove the dependency of the Zipr statistics
class on Zipr options. Also, re-enable printing
of the stats at the end of the Zipr.

update for RNG seeding to use pid() instead of time() due to parallel copies getting the same seed accidentally

Test for some of the new type-specific
actions that can be performed on types.

Updates to get file-sizes out of zipr under control.  should now work with .exe's, .so's, and executable .so's.  note:  assumes that if entry_point==.text's address that the .so is not-executable (because the .text is the compiler's default entry point, indiciating the user did not specify an entry point.)  May not be a rock solid assumption, but it should be OK for 99.99% of all programs

1. Did not properly detect missing required values
for arguments

2. Could not handle parsing without a error and
warn IO handle.

Broken version being committed, so will can commit his broken version so we can both finish our hacking without dealing with repo conflicts.

Updated to write zipr.map for machine/human reading of the 'map' between instructions and their final location.

1. Add support for plugins to call certain
Zipr_t functions. This meant making a class
that is Zipr_t and then converting the actual
implementation to a subclass (ZiprImpl_t). The
Zipr_t goes in the zipr_sdk.

2. Ask all plugins about each instruction
and determine whether there are any takers
for plopping.

3. When plopping an instruction, consider
whether or not a plugin asked to do the plopping.
If so, let it do the work.

updated zipr to use add_segment instead of move_segheaders for corredtness and size effeciency for CGC

That's a mouthful: No replop is the default and
the -r option will make it so that dollops _are_ replopped.

that start/end on the same byte are actually
1-byte big. Update tests to confirm this.

calling RemoveFreeRange().