Add support for building custom Search Pipelines

[tanaysoni]

Related issue: #544

[tholor]

Very nice!
This will be the basis for many exciting changes in the future :)

[tholor]

Just adding a small usage example here for completeness:

from haystack.pipeline import DocumentSearchPipeline, ExtractiveQAPipeline, Pipeline, JoinDocuments
from haystack.document_store.elasticsearch import ElasticsearchDocumentStore
from haystack.reader.farm import FARMReader
from haystack.retriever.sparse import ElasticsearchRetriever

# Building blocks (= Nodes)
document_store = ElasticsearchDocumentStore(host="localhost", username="", password="", index="document")
retriever = ElasticsearchRetriever(document_store=document_store)
reader = FARMReader(model_name_or_path="deepset/roberta-base-squad2", use_gpu=False)

# Combine via default pipeline (here: QA)
qa_pipe = ExtractiveQAPipeline(reader=reader, retriever=retriever)
res = qa_pipe.run(question="Who is the father of Sansa Stark?", top_k_retriever=2, top_k_reader=5)
print(res)
qa_pipe.draw()

# Combine via default pipeline (here: Document Retrieval)
doc_pipe = DocumentSearchPipeline(retriever=retriever)
res = doc_pipe.run(question="Who is the father of Sansa Stark?", top_k_retriever=2)
print(res)

# Or build your own custom pipeline to model complex search routes.
# Choose existing components or build your own. Stick them together to a DAG
p = Pipeline()
p.add_node(component=retriever, name="ESRetriever1", inputs=["Query"])
p.add_node(component=retriever, name="ESRetriever2", inputs=["Query"])
p.add_node(component=JoinDocuments(join_mode="concatenate"), name="JoinResults", inputs=["ESRetriever1", "ESRetriever2"])

In future PRs, we should improve the execution under the hood, provide more standard "components" and add utility methods to import / export to yml