Previous implementation of C interface passed raw pointers
up to user and it was unclear and inconsistent who owned the pointer
leading to bugs. The previous implementation lead to creation
of many copy constructors, many of which had bugs.

Now we pass opaque pointers to C which the user is *always* responsible for
freeing.

Remove broken copy constructors.

Update C tests so that all pass valgrind memory leak checks.

it breaks windows and some of Xdmf's own tests

Revert "removed a PARENT_SCOPE that was unessesary, added some missing includes, fixed cxx11 typecasting"
This reverts commit 583c14de.

Revert "changes to fix windows build"
This reverts commit 397e235f.

Revert "Updates to reduce library files needed for linking"
This reverts commit ff194c1f.

Revert "adjusting code away from regex to handle Apple systems"
This reverts commit 05124fbf.

Revert "Fixed typo"
This reverts commit bba7984f.

Revert "std::string compatibility"
This reverts commit 24ea357e.

Revert "fixes to handle boost not being available"
This reverts commit b21b735a.

Revert "removed print statement"
This reverts commit 44d1df0e.

Revert "Fixed typo in PushBack change"
This reverts commit 61df7337.

Revert "fixes for Windows Compilation"
This reverts commit 92313779.

Revert "Changes to remove boost dependency and add Cxx11 as an alternate"
This reverts commit f5300338.

Changes from ARL's Repo (C Wrappers, DSM rework, Custom Types, Templates, GridControllers, TIFF integration, XdmfAggregate)

Add HDF5 tree writer test.

Eliminate getRealPath calls which are expensive and not necessary in many cases
Parse XML node content only when necessary (parsing arrays)
Fix cases where copies were made of strings instead of using references
Use "map.insert()" rather than "map[] =" when adding properties to 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boost::shared_ptr and boost::shared_dynamic_cast have been replaced
by shared_ptr and shared_dynamic_cast.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: 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: Commit Ken Renard's patches modifying populateItem and XdmfCoreReader to support some xpath functionality for backwards compatibility.

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

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.

to write out the Xdmf graph.  Modify some random include statements for style.

ENH: Add pure virtual getItemTag method to XdmfItem working to return the tag (same as in XML terminology) for the specific item.

ENH: Add XdmfItemProperty to genericize properties attached to XdmfItems.  Add a pure virtual getItemProperties() method
to XdmfItem to force children to implement a way to retrieve properties.

ENH: Create a generic XdmfVisitor class and move writing operations to XdmfWriter.  Incorporate Loki library to perform asymmetric visitor operation (allows customizable behavior for eventual Xdmf / NetDMF library split).  Modify Loki
Visitor to work happily with boost::shared_ptr and modify swig wrapping to properly handle boost::shared_ptr inheritance.

ENH: Rename write() operation to accept() which is more consistent for the Visitor pattern.  Also, we can
do much more to our Xdmf graph structure than just write it using Visitor operations.

unnecessarily duplicated functionality.

ENH: Split visitor operation into a visit and a traverse operation so that logic of how to traverse the xdmf graph is kept with xdmf objects rather than the visitor.  The visitor calls this function rather than relying on its own logic.

	Add XdmfArray to store values --- modify XdmfDataItem to contain an XdmfArray.
	Modify Tests to use new XdmfArray.
	Modify XdmfVisitor to use new XdmfArray.
	Add Python Tests and move Cxx tests to their own directory.

ENH: Make XdmfAttribute, XdmfTopology, XdmfGeometry inherit from XdmfDataItem.  Instead of containing a data item, each is a dataitem of a specific type.  This makes setting data items more user friendly --- e.g. once you set the number of elements and set the topology type you don't have to set the number of elements from the dataitem again.  This still needs to be figured out for composite dataitems for things like sets.

ENH: Add XdmfAttributeCenter to XdmfAttribute to specify how the attribute values are attached to the grid.

ENH:  Replace old way of creating static instances of XdmfTopologyType, XdmfGeometryType, and storing them by reference in their respective topologies.  These now work by providing a specific constructor method that returns specific instances of the objects populated with correct values.  This was changed because there are some type options in Xdmf where static instances don't make sense (e.g. XdmfTopologyType Polygon can have many different values for nodesPerElement) but we still need to restrict construction of these types with specific values.  Each parent item merely keeps a copy of a topology type.  I considered doing this with smart pointers, but I believe it to be unnecessarily complicated, especially when direct comparison of topology types is needed.  I am considering separating these types into separate classes.

Added XdmfAttribute implementation and began populating XdmfVisitor to write out XML.