Project restructuring

Should I go for more magic, or less?
### Ideas:

``` python
from simplegexf import Gexf, Graph, Edge

class MyGraph(simplegexf.Graph):
defaultedgetype='directed'

node_attributes = [
('name', 'string'),
('description', 'string'),
('weight', 'integer'),
]

edge_attributes = [
('rel_type', 'string'),
]

gexf = simplegexf.Gexf('/path/to/file.gexf')

try:
graph = gexf.graphs[0]
except KeyError:
graph = MyGraph()
gexf.graphs.append(graph)

tag_list = [
{
'id': 'c6fa996d-8593-484d-b067-69bc828bbeba',
'name': 'Test Tag',
'description': "Fairy penguins citylink, east brunswick club trams",
'weight': 13,
'parents': [], # List of UUIDs
}
...
]

# Create Nodes:
for tag in tag_list:
try:
# graph.nodes behaves like an OrderedDict, and nodes get mapped by id,
# so it will throw `KeyError` if the key does not exist:
node = graph.nodes[tag['id']]
# or, graph.nodes should behave like a list, and we use a method to assist:
node = graph.nodes.get(tag['id']) # As in get_by_id()
except KeyError:
node = Node(tag['id'])
# graph.nodes behaves like an OrderedDict, but also like a list,
# so we can append.
graph.nodes.append(node)
# I'm not sure about this. Maybe it's best if graph.nodes behaves
# like a list.

# node.attributes is an OrderedDict, so .update takes a list of tuples:
node.attributes.update([
('name', tag['name']),
('description', tag['description']),
('weight', tag['weight']),
])

# Create Edges:
for tag in tag_list:
for parent_id in tag.get('parents', []):
try:
edge = graph.edges.get(parent_id, tag['id'])
except KeyError:
edge = Edge(parent_id, tag['id'])
graph.edges.push(edge)

edge.attributes['rel_type'] = 'parents'
```
### Less Python Magic:

``` python
from simplegexf import Gexf

gexf = simplegexf.Gexf('/path/to/file.gexf')

try:
graph = gexf.graphs[0]
except KeyError:
graph = gexf.create_graph()

graph.define_attributes([
('name', 'string'),
('description', 'string'),
('weight', 'integer'),
])

graph.define_attributes([
('rel_type', 'string'),
], _class='edge')

tag_list = [
{
'id': 'c6fa996d-8593-484d-b067-69bc828bbeba',
'name': 'Test Tag',
'description': "Fairy penguins citylink, east brunswick club trams",
'weight': 13,
'parents': [], # List of UUIDs
}
...
]

# Create Nodes:
for tag in tag_list:
try:
node = graph.get_node(tag['id'])
except KeyError:
node = graph.create_node(tag['id'])

# even...
node = graph.get_or_create_node(tag['id'])

node.attributes.update([
('name', tag['name']),
('description', tag['description']),
('weight', tag['weight']),
])

# Create Edges:
for tag in tag_list:
for parent_id in tag.get('parents', []):
try:
edge = graph.get_edge(parent_id, tag['id'])
except KeyError:
edge = graph.create_edge(parent_id, tag['id'])

# even...
edge = graph.get_or_create_edge(parent_id, tag['id'])

edge.attributes['rel_type'] = 'parents'
```

No files or tests are named. Start by reviewing the existing simplegexf public API and compare it with the two example designs; clarify whether the project should favor class-based or explicit helper methods before defining a scoped restructuring plan and acceptance criteria.

Written by the indexing model from the issue text.