documentation for recognition of multiple wakewords

Hello, 
can you please add documentation on how to detect multiple wakewords? Is it preferable to train a separate model for each wakeword  or to train a single model supporting several wakewords?

Also do the "spot" and "test" commands support multi wakeword detection? 
