Vector and Index Graph API improvements

[miratepuffin]

For search APIs

#Currently we have to generate an IndexGraph via 
index = financial_graph.index()
#Instead I should be able to just call search_edges/search_nodes and internally if we have an index use that, if not create one
index.search_edges("from:Polux OR to:Hilux purpose:Payment")[3]

For vector APIs

#Current function call for vectorising - suggestion would be to change node_document and edge_document to functions which take node/edge and return List[Documents]
financial_graph.vectorise(lambda x: embeddings.embed_documents(x),node_document="document",edge_document="document",cache="embedding-cache-notebook",overwrite_cache=True)

Most functions below will error if the graph has hes to be vectorised - i.e. calling the function above

#Create a new empty selection context
selection = financial_graph.vector_context()
#Create context with 10 nodes (based on top 10 most similar documents, add all documents for each node)
selection2 = financial_graph.nodes_by_similarity("blah",10)
#Create context with 10 edges (based on top 10 most similar documents, add all documents for each edge)
selection3 = financial_graph.edges_by_similarity("blah",10)
#Create context with 10 nodes/edges (based on top 10 most similar documents, add all documents for each node/edge)
selection3 = financial_graph.entities_by_similarity("blah",10)
#Create context with 10 documents (based on top 10 most similar documents)
selection3 = financial_graph.documents_by_similarity("blah",10)

#Combine selections based on typical joins
selection.union(selection_2)
selection.intersection(selection_3)

#Add edges based on tantivty index search
index_search = financial_graph.search_edges("from:Polux OR to:Hilux purpose:Payment")
selection.add_edges(index_search)
#Can do same with node search/selction of nodes
selection.add_nodes(vec_of_nodes)

doc_list = []         #or .edges  
for doc in financial_graph.nodes.documents:
    #some filtering logic
    doc_list.append(doc)
# complex logic to decide specific documents to add
selection.add_documents(doc_list)

#top 10 ranked neighbours of nodes within the selection 
selection.expand_nodes_by_similarity("blah blah blah",10)
#top 10 ranked edges of nodes within the selection
selection.expand_edges_by_similarity("blah blah blah",10)
#top 10 ranked nodes/edges of nodes within the selection
selection.expand_by_similarity("blah blah blah",10)

selection.nodes() # -> Node objects from graph
selection.edges() # -> Edge objects from graph
selection.documents() # -> Document object which has reference to the content, the associated entity, the raw embedding 
selection.token_size() # -> how many tokens this context would be 

#Document attribute
document.entity
document.embedding
document.token_size # -> how many tokens the document is 
document.content
document.lifetime # -> the timerange for which this doc is valid