Add a fetcher manipulator that will let you use mutiple CSV files as input

[mjordan]

For large, complex migrations (for example), CSV input data might come from several sources, e.g., info about files might come from a separate source than descriptive metadata. It should be possible to have MIK combine multiple CSV files based on common record_key values. Or, it might be useful to allow for alternative shared keys, which could be specified in the manipulator's configuration.

I am currently involved with a migration where this would be useful.