> Masses attract because they are attracted to eachother is not a theory of gravity, it's a restatement of the givens.Newton's theory of gravity had been dominating for a several centuries. It states literally this, masses attract because they are attracted. Despite of this it is a very valuable theory even now.
> Theories explain observations, and predict what ought to be observed beyond the data used in their formulation.
A very good theories manage to combine good explanations with a predictive power. But there are countless examples of theories which could give either explanation or predictive power. Moreover there are theories which manage neither explanations nor predictions, but something else. For example they can be used as an data structure for storing and retrieving facts: Carl Linnaeus devised the Linnean taxonomy, which directed biological research for a long time, even there were nor explanations nor predictions.
I'm not sure though, maybe I'm stretching the term "theory" a little? But if we asked functions of Linnean taxonomy in times before Darwin, then we'd find that it's functions were functions of a theory: to collect and to classify facts and to drive research.