Perhaps you must "unzip" a little bit your question, because it is very wide in scope.
You are alluding at least at three different and very interesting issues :
historical interpretation of scientific theories and controversies
current philosophy of science debates
historical (meta-)interpretation of the interpretation of history of science by a distinguished scientist, historian and philosopher of science like Pierre Duhem.
Here are some personal comments:
(1) the preface by Osiander to Copernicus' book, that is a typical example of the motto "saving the appearences" must be read in context; I think that it was more a way of avoiding troubles with "traditionalists" (i.e.peripatetics and theologians - see the quotation form Osiander's letters in Edward Rosen (editor), Three Copernican Treatises (3rd ed - 1971), page 23) regarding Copernicus' theory of the moving earth, than a real "philosophical" position.
(2) having said that, it is quite certain that Copernicus was a "realist" : he believed in the moving earth. Scientific realism means - in a few words - the "philosphical" assumption that our current scientific laws and theories are faithful descriptions of the world's facts and structure, i.e. of the reality "out there".
Newton's law of gravitation describes the behaviour of moving bodies under the action of a central force called gravitation that "pull them" according to a definite proportion... full stop.
According to this philosophical position, quanta are "out there": Higgs' boson is "out there", and all the stuff of "strage things" invented by physicist are "out there" (also strings ? ...)
An interesting argument of a distinguished contemporary philosopher, Ian Hacking, in his book Representing and Intervening, can be synthetized as follows : the strange "actors" of the subatomic world are difficult to imagine, but we may "interact" with them (with complex macìhines like the Large Hadron Collider of CERN, ...); so, if we can "push and pull" them, they must exists.
(3) but at the same time, scientific theories has changed during time: Lavoisier phlogiston has disappeared; what about aether of pre-einsteinian physiscs ? So it seems more reasonable to think that our scientifc theories are "useful models" that we can use profitably to calculate and manipulate facts, independently of their real ability to grasp the reality "out there". They are used more to "save the appearences", i.e. to calculate with a reasonable precision how the planets will move, more than trying to "discover" the hidden mechanism of their motions.
(iv) Pierre Duhem, end of XIX century, made a wonderful work in rediscovering the traces of medieval (pre-galilean) science and, at the same time, was driven by his contribution to the transformation of physiscs in his epoch, to "re-evaluate" an anti-realistic interpretation of science.