given a biconnected symmetric graph with weighted edges,

what is the algorithmic complexity of determining a set of pairwise edge-disjoint cycles with maximal sum of edge weights *if there are no other constraints* besides edge-disjointness of the cycles and maximal weightsum of their edges?

Determing such a set of cycles is a stepping stone in an algorithm for determining a heaviest euler tour in complete symmetric graphs with $n=2k$ vertices (which isn't eulerian), which in turn would yield an improved heuristic for the non-eulerian windy postman problem.