memory leak in active particles

[Matthew Turk]

Hi,

For the issue brought up at the workshop, here's the dataset I've been using:

http://use.yt/upload/c3c392ae

My valgrind command:

mpirun -np 2 valgrind --tool=memcheck --leak-check=full
--error-limit=no ../../../src/enzo/enzo.exe -d -r DD0015/DD0015

It'll show up after a few cycles. Some of the errors look like:

http://paste.yt-project.org/show/5913/

As I mentioned in person, I *believe* that what's happening is
particles that are not being deleted from processor N get overwritten
when the buffer sent from processor N to itself gets Unpack'd. I may
be wrong. But note that the valgrind error report only shows where
the memory was allocated, not where the pointers got overwritten.

-Matt