Plotting doesn't work on a clean install

[LukeWeidenwalker]

Currently, plotting doesn't work out of the box for two reasons:

1. igraph requires the pycairo or cairocffi libraries to be installed, these are not specified in setup.cfg
2. Upon installing cairocffi, calling parsed_pg.plot() results in the exception quoted further below. Process graph also below.

I've been able to circumvent this problem quite easily by using the networkx library instead.
Happy to raise a PR if we're okay with replacing igraph.

@clausmichele I don't know who uses this library / who is responsible for it these days, could you please tag any folks you think should have a say in this? Much appreciated :)

---

---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
File /opt/conda/lib/python3.9/site-packages/IPython/core/formatters.py:343, in BaseFormatter.__call__(self, obj)
    341     method = get_real_method(obj, self.print_method)
    342     if method is not None:
--> 343         return method()
    344     return None
    345 else:

File /opt/conda/lib/python3.9/site-packages/igraph/drawing/__init__.py:367, in Plot._repr_svg_(self)
    365 context = cairo.Context(surface)
    366 # Plot the graph on this context
--> 367 self.redraw(context)
    368 # No idea why this is needed but python crashes without
    369 context.show_page()

File /opt/conda/lib/python3.9/site-packages/igraph/drawing/__init__.py:276, in Plot.redraw(self, context)
    274 else:
    275     ctx.save()
--> 276 plotter(ctx, bbox, palette, *args, **kwds)
    277 if opacity < 1.0:
    278     ctx.pop_group_to_source()

File /opt/conda/lib/python3.9/site-packages/igraph/__init__.py:4166, in Graph.__plot__(self, context, bbox, palette, *args, **kwds)
   4164     del kwds["drawer_factory"]
   4165 drawer = drawer_factory(context, bbox)
-> 4166 drawer.draw(self, palette, *args, **kwds)

File /opt/conda/lib/python3.9/site-packages/igraph/drawing/graph.py:489, in DefaultGraphDrawer.draw(self, graph, palette, *args, **kwds)
    487 src, dest = edge.tuple
    488 src_vertex, dest_vertex = vertex_builder[src], vertex_builder[dest]
--> 489 (x, y), (halign, valign) = edge_drawer.get_label_position(
    490     edge, src_vertex, dest_vertex
    491 )
    493 # Measure the text
    494 _, yb, w, h, _, _ = label_drawer.text_extents()

File /opt/conda/lib/python3.9/site-packages/igraph/drawing/edge.py:159, in AbstractEdgeDrawer.get_label_position(self, edge, src_vertex, dest_vertex)
    155     angle = None
    158 # Determine the midpoint
--> 159 if edge.curved:
    160     (x1, y1), (x2, y2) = src_vertex.position, dest_vertex.position
    161     aux1, aux2 = get_bezier_control_points_for_curved_edge(x1, y1, x2, y2, edge.curved)

AttributeError: 'igraph.Edge' object has no attribute 'curved'

This occurred on the following process graph I was experimenting with:

{'process_graph': {'loadcollection1': {'process_id': 'load_collection',
   'arguments': {'bands': ['B02', 'B03', 'B04', 'B08'],
    'id': 'boa_sentinel_2',
    'spatial_extent': {'east': 9.992539570642537,
     'north': 47.73255902176489,
     'south': 46.82379186182021,
     'west': 8.259091264204697},
    'temporal_extent': ['2018-05-01', '2018-09-01']}},
  'reducedimension1': {'process_id': 'reduce_dimension',
   'arguments': {'data': {'from_node': 'loadcollection1'},
    'dimension': 't',
    'reducer': {'process_graph': {'median1': {'process_id': 'median',
       'arguments': {'data': {'from_parameter': 'data'}},
       'result': True}}}}},
  'loadvectorcube1': {'process_id': 'load_vector_cube',
   'arguments': {'URL': 'https://raw.githubusercontent.com/LukeWeidenwalker/SRR3_clustered_geojsons/master/AOI_clustered_distance_2/cluster_3.geojson'}},
  'aggregatespatial1': {'process_id': 'aggregate_spatial',
   'arguments': {'data': {'from_node': 'reducedimension1'},
    'geometries': {'from_node': 'loadvectorcube1'},
    'reducer': 'mean',
    'target_dimension': 'result'}},
  'fitregrrandomforest1': {'process_id': 'fit_regr_random_forest',
   'arguments': {'data': {'from_node': 'aggregatespatial1'},
    'max_variables': None,
    'num_trees': 100,
    'predictors_vars': ['B02', 'B03', 'B04', 'B08'],
    'seed': 0,
    'target_var': 'target_canopy_cover'}},
  'savemlmodel1': {'process_id': 'save_ml_model',
   'arguments': {'model': {'from_node': 'fitregrrandomforest1'}},
   'result': True}}}