This proposes several new traits that are useful in manipulating existing graphs.

pub trait GraphMap {
	type Output;

	fn map<'a, F, G, N, E>(&self, nodes: F, edges: G) -> Self::Output<N, E>;
}

// mutable access could be modelled over `for node in g.nodes_mut() { node = ... }`

pub trait GraphFilterMap {
	type Output;

	fn filter_map<'a, F, G, N, E>(&self, nodes: F, edges: G) -> Self::Output<N, E>;
}

pub trait GraphRetain {
	fn retain_nodes<F>(&mut self, visit: F);
	fn retain_edges<F>(&mut self, visit: F);
}

These are modeled after Graph (and StableGraph) as they are currently the most fully featured.

We might want to think about the following:

• edge_connecting()
• contains_edge()
• find_edge()
• find_edge_undirected()
• externals()
• node_weights()
• node_weights_mut()
• neighbors_undirected()
• reverse()
• Entry API

On Graph, there are the following duplicates:

• node_indices() -> IntoNodeIdentifiers (deprecate)
• node_weight() -> Trait Rework: Graph fundamentals #556 GraphAccess (deprecate)
• node_weight_mut() -> Trait Rework: Graph fundamentals #556 GraphAccess (deprecate)
• edge_weight() -> Trait Rework: Graph fundamentals #556 GraphAccess (deprecate)
• edge_weight_mut() -> Trait Rework: Graph fundamentals #556 GraphAccess (deprecate)
• neighbors() -> IntoNeighbors (deprecate)
• neighbors_undirected() -> IntoNeighorsUndirected (deprecate)