search = vectordb.similarity_search(" ")

this returns only 4 docs.

is there a way to return all documents from vectordb?

isn't this what you want since the purpose is to get a summary of a corpus which comprises of all texts?