Gravar-mail: Contest models highlight inherent inefficiencies of scientific funding competitions