allow for missing metadata fields in streams-persist-hdfs

Currently streams-persist-hdfs writer creates (and reader expects) exactly four columns. this could be made much more flexible without too much effort.

Update reader to support additional use cases:
a) file paths containing one json document per line
b) file paths containing just id and json on each line,
c) file paths containing id timestamp and json document on each line

Update writer support
a) ids only
b) ids and timestamp only
c) ids timestamp and json only