Just input file into partitioner with new number of processes. The file
is unpartitioned in memory and repartitioned.

Use -u flag to unpartition a previously partitioned mesh.

Previously, each time a dataset was written to hdf5, the file was opened
and closed. This was implemented because inside the heavy data writer
we have no idea whether another dataset will be written, so we don't know
whether to keep the file open. After profiling large xdmf writes it was
determined that closing the file appeared to be a bottleneck so hooks were
added to the heavy data writer to open the file and keep it open while writing.
This is optional, but was incorporated into several higher level classes
like XdmfWriter where we know we can leave a file open until we finish writing
the XML. For our test problem we saw a 7 time performance gain.

Unpartition method takes an XdmfGridCollection of unstructured grids
and zips them back together into a single XdmfUnstructuredGrid. It uses
GlobalNodeIds to perform the unpartitioning.

Also added some functionality to the partitioner so that it will properly
partition set attributes of types other than scalar.

The new version of metis uses a different format to supply connectivity
information to the partitioning algorithm. The new format allows generic
element types to be passed (as well as mixed topology formats). These changes
expand the number of element types we can partition in metis (this was tested
on high-order hexahedron, linear hexahedron, and polyline topologies)

Modify XdmfTopologyConverter and XdmfPartitioner to work with new
topology types.

partitioned incorrectly by the partitioner.

more than one map to be attached.

Add additional attribute to XdmfMap, which is the name of the map.

Using statements have been added to XdmfSharedPtr.hpp. All used of
boost::shared_ptr and boost::shared_dynamic_cast have been replaced
by shared_ptr and shared_dynamic_cast. Slight modification to XdmfCore.i
to get swig wrapping working again.

Hopefully this will become a standard style for Xdmf development.
All spaces - 80 character line wrapping.

ENH: Rename all subclasses of XdmfGrid to more syntactically friendly names: i.e. XdmfGridUnstructured to XdmfUnstructuredGrid.

ENH: Added XdmfGridUnstructured and turned XdmfGrid into an abstract base class for all grids (both structured and unstructured).  Modified a bunch of utils and other classes
to fit this new hierarchy.

ENH: Change some method names in XdmfArray to be clearer and less verbose.  Modify files for documentation clarity.  Modify some minor things to prevent -Wall warnings in gcc.  Add python install rules and rpathing.

ENH: Begin adding install rules.  Modify bug in XdmfWriter that resulted in XdmfArrays not being written in certain cases.

ENH: Modify several utils to take XdmfHDF5Controller as an argument so that they can write data to hdf5 as generated rather than keeping it all in memory.  This is an optional argument.

ENH: Add XdmfExodusReader for reading ExodusII files into Xdmf.  Add ability for XdmfArray::initialize() to take a size value to initialize the array to.

ENH: Rethink XdmfWriter / XdmfCoreWriter distinction and move XdmfSet to XdmfArray conversion in XdmfSet's traverse method.  If conversion is unneeded then do not traverse XdmfSet.

ENH: Make XdmfGridCollection inherit from XdmfGrid and XdmfDomain, since it is really a unity of the two classes.  Modify some details in ItemFactory / Reader.

ENH: Rename get/set ItemProperties for conciseness.  e.g. getAttributeCenter ---> getCenter .  This follows
what is done for XdmfArray (getType).

ENH: Begin implementing XdmfTopologyConverter.  Fix typo in XdmfArray that incorrectly initialized UInt32() type.

ENH: Modify ItemProperties to use shared pointers like the rest of Xdmf.  Modify swig wrapping to correctly
handle both non const and const shared pointers.