Induced subgraphs

We define `schematic_graph[I] = [nodes(schematic_graph)[i] for i in I]` while `metagraph[I]` gives the induced subgraph on nodes indexed by `I` (that is, a graph made up of all of those nodes and any edges between them). This would be really useful! It would also be breaking. We can still implement an `induced_subgraph` method that does this on the internal graph and reconstructs the `SchematicGraph` appropriately.

Start by locating the SchematicGraph type and its internal graph, then compare the existing schematic_graph[I] behavior with metagraph[I]. Trace how a new SchematicGraph is reconstructed from selected nodes and edges. Done means an induced_subgraph method returns the induced graph on the requested node indices without changing the existing indexing behavior.

Written by the indexing model from the issue text.