#153 Improve conformer-related API

[Konstantin Tokarev]

Currently, "conformers" in OB can be: a) list of real conformers; b) list of geometry optimization steps; c) if relaxed scan is performed, sequential compozition of lists of all geometry optimization steps. However, it would be useful to distinguish a) and b) (e.g., in latter case conversion to Hartree units are more desirable in external application, in former - not). Case "c" also needs to be distinguished - minimums of every optimization cycle are "real" conformers in a) sense, and lists of every optimization need to be separated to simplify life of users