Add snips STT

[Sispheor]

Just a feature we could propose if the integration is possible.
Here is a doc. the STT part is actually called ASR (Automatic Speech Recognition).
From my understanding the idea would be to send a MQTT message to the ASR part and subscribe to a MQTT thread to get the text.
We cannot caught the audio by ourself and then send him to SNIP engine. Snip seems in change of capturing the audio. And so, Kalliope and snip will need to be on the same device.

[Sispheor]

Seems that snips is based on Kaldi.
Maybe help the speeh reco lib to integrate it would be better then. I can see that a PR to do this has been proposed a long time ago. Not merged yet.