QAT: Better handling of optimizer and of creation of fp32 weights copy (#399)

* Create float copy such that the actual tensor being learned stays
  the same
* This way the optimizer doesn't have to be re-created, just need to
  add parameter groups if algo requires it (e.g. PACT)
* This also means we don't care about pre-existing parameter groups,
  as opposed to the previous implementation which ASSUMED a single
  existing group