Oussama Danba caught a bug in how we were handing a test related to
ciphertext encoding. Not a security issue as far as we can tell, but
certainly not the correct behavior.
More detail here: e0ab9525f1
This commit also switches ntruhrss701/avx2/poly_S3_inv to the
Bernstein--Yang code, also slightly simplifies the poly_Rq_to_S3
routine.
There are schemes, like SABER (#192) that have more than one principal
submitter. Consistency warrants that we turn it into a list for all
schemes and don't do something with allowing either a str or a list:
that would just be very annoying to parse.
Closes#194
