OPEN ISSUES:
------------

  
Issue: Segments in pmesh that are solely used to separate regions 
(e.g. for different resolution) will appear as an *internal* 
boundary in the resulting domain even though it may not have been intended to 
have a boundary condition applied.
Background: Genuine internal boundaries are created the same way as 
external boundaries through the 'boundary' dictionary mapping triangle_id 
and edge number to a tag.
When a mesh is created, this dictionary will override the neighbour structure 
and thereby enforce the internal boundary status.
This is all very well for 'True' internal boundaries.
However, once a segment belongs to a boundary (internal or external) 
it must be bound to a boundary condition object in order to supply 
values for the flux computations. Hence there is no way of allowing 
normal flow between neighbouring triangles in the case where they are 
separated by an (unintended) internal boundary.
An older version of pyvolution allowed None as a boundary object which 
probably meant that a zero dirichlet condition was imposed. 
Not sure about this, though.
Importance: High
Suggested Action: Have pmesh ignore segments that aren't being tagged
and that way they are exclusively used to to define regions 
(e.g. for variable resolution).
Status: Unresolved