Minor documentation issue with training model (train.sh) on new dataset.

[al-alamin]

In the documentation, it suggests adding '-o' to provide a new name for the newly trained model.
sh train.sh -i train.csv [-d csv_delimiter] [-g] [-c chunk-size] [-j jobs-number] [-o model-name]

But on the train.sh, file necessary code is not implemented properly. So, in its current form, it is required to use -m instead of -o during training to save the model with a user-provided name.

sh train.sh -i train.csv [-d csv_delimiter] [-g] [-c chunk-size] [-j jobs-number] [-m model-name]