When a node in the graph is deleted, it can sometimes cause confusion with the view provided by the `Graph.adjacentNodes` function. This happens because the view might have been created before the node was removed.

The removeNode function in `StandardMutableGraph` (which calls `StandardMutableValueGraph.removeNode`) meant to eliminate the connections related to the node. However, the `{Directed,Undirected}GraphConnections.adjacentNodeValues` associated with the to-be-removed node might still be referred to by the view constructed by `Graph.adjacentNodes` (which calls `StandardValueGraph.adjacentNodes`).

So, this can result in indirect references to the values of adjacent nodes (from the perspective of the node that was removed), even though they've been deleted. This makes the view less transparent, because it's showing information that isn't accurate anymore after the node has been removed.

The solution is to cleanup the associated `GraphConnections` before finally removing the node from the graph. Also it is now better aligned to the implementation of `StandardMutableNetwork.removeNode` reusing the same mechanism of copying the view to an immutable collection to avoid modifying the underlying view while iterating over it.

Fixed review finding: the approach was borrowed from `StandardMutableNetwork.removeNode` where elements are copied to `ImmutableList`. Making the immutable collection list here too, so it is safer as no deduplication or reordering might happen.