Semiring Refactoring
Moved semiringType to own header and used it as a tag for the semiring class + implemented torchs own matrix multiplication for classic Semirings (Lead to a speed up of about 2,54 on my system for classic semirings)
cleaning up