read_list returns int value on missing object keys

[rsdoiel]

If you request a list of keys and some are not found in the list passed to dataset.read_list(c_name, keys) then the value returned in an integer. This is not helpful. It should be recording an error in generating the list of objects but return the list for the ones present.

Encounter this when using dataset.read_list() to generated a list of citations and I happened to pass the wrong values for keys.

Also notice I have some really clumsy handling of encoding/decoding values to and from strings in dataset.py. I can simplify that code by checking if isinstance(value, bytes) then doing value = value.decode('utf-8') per the Python idiom to get things to a str before handing to json.loads().

[rsdoiel]

Even when passing in correct values this fails. When dataset was refactored the Read func only takes one key. libdataset.py needs to implement read_list in Python from successive calls to dataset.read(). Or I need to remove read_list.