The purpose of this change is to share this trait with other crates,
such as the forth-coming "semiring" crate that will be responsible for
handling some simple semiring operations as well as the querying, in
my plans.

Not bug deals but adjustments of details.

* chain/src/item/default/splone.rs: A function for finding the node to
  plant used to ignore the condition that the children should match.
  Now this is fixed.

Some bugs are fixed:

1. If a non-terminal expansion can be reduced immediately, previously
an extra node would be created that had no parents.  Now this strange
behaviour is corrected.

2. When performing reductions, a leaf non-terminal node would
previously be regarded as completed.  Now we will first try to
complete that node, and then determine if the completion is
successful, and finally determine the completedness according to the
result.

Of course some more tests are still pending, before I can confirm that
no more bugs lurk around.

* chain/src/item/default/printer.lldb:
* chain/src/item/default/printer.py: These are for experimenting with
  debugger supports.

* chain/src/item/default/mod.rs:
* graph/src/labelled/binary.rs:
* graph/src/labelled/double.rs:
* graph/src/lib.rs: If we set the option "ordering" to be "out" in the
  declaration of nodes at the beginning, then GraphViz will not change
  the order of children out of nodes.  This is much better looking in
  my opinion.

* INSTALL: make insists in changing this file, so let it be.

This bumping of version is insignificant.  I just find it notable that
I seem to finally obtain a version without trivial bugs.  Hooray!

* chain/src/item/default/splone.rs: Previouslt the function
  `split_node` used to split the parents of splitted nodes by an ugly
  logic.  Now that is moved into a dedicated function, which properly
  handles the splitting of parents, including the case when the new
  node is open whereas the old node is closed, in which situation we
  ought to put the new node under the opened node only, as a closed
  node cannot contain an open node as a child by definition.

The function `set_pos` is kind of subtle and its behaviour needs a
unit test so that we can be sure that it does not accidentally set the
ending positions in a careless manner.

Previously when generating a fragment of the forest corresponding to
the expansion of a non-terminal by a terminal, we incorrectly set the
end of every node within it to be one plus the start, if the expansion
happens due to a reduction.

Now this mistake is fixed and the ending positions are correctly set.