Support draft-ietf-softwire-yang-16

I have changed the way model translation works to map the lwaftr to a single bind-instance (the device instances are now invisible from the ietf-softwire-br model). I think this better reflects the semantics on the model, in which binding tables are per-instance.
I have also extended the test cases for the translator to cover more cases.
I have added a change to the snabb-softwire-v2 schema: 7a17eba (declare softwire/b4-ipv6 unique). I think this is required because this leaf is used as the key in the ietf-softwire-br schema and hence must be unique in order for the binding table to be translatable without loosing entries. (I am somewhat suspicious of the performance implications this has for large binding tables.) Definitely needs review Cc @wingo

Igalia / snabb