Fuzzy clustering of distribution-valued data using an adaptive L2 Wasserstein distance