images that only have ignore boxes when you load into mmod_rects like it should
have been.

* remove unused variable

* modify variable type from int to size_t

* fix previous delete, we need to call chinese_whispers here but we don't need its result now

* cont layer: allow to set the number of filters at runtime
(implementation essentially copied from con layer)

* Document the new constructor cont_(num_con_outputs o), and the function set_num_filters

sizes.  It would pick a bad size in some cases.

if small images are given as input.

In particular, these new functions don't need to be inside the face
recognition class.  So I moved them out.  I also fixed many incorrect
copy/pasted comments and clarified parts of the example code.