When oem is set to 1 (LSTM_ONLY) the LSTM-only data should be loaded, however this is not happening--the LSTM + Legacy data is still being loaded. This can be verified by looking at file sizes--the integerized LSTM-only data is only ~2MB for most languages.

In addition to fixing this issue, new unit tests should be added to confirm that the correct data is being downloaded. The unit tests we have hard-code what language data is used, which effectively bypass the language data from being tested. As the vast majority of users presumably use the default arguments, it is important that this be tested.