Read Jake's interesting post here.
This is an excellent response, and I find myself agreeing with you, though I still have trouble with the word "craft." For me it is too heavily weighted with both New Critical assumptions and common workshop practice to be useful to my own thinking. If we follow your example, and discern the principles that make a "good" O'Hara poem, and then, on these principles, determine which of O'Hara's poems are well-crafted, we're still essentially saying that a well-crafted poem is one we like--
The poems I like (the "good" poems) contain elements X, Y, Z, which may be different than elements A, B, & C, which are the most prized elements in Po Land (or Bizarro Po Land, depending on how you want to stack your Xs, Ys, and Zs). If Poet Alef hits X,Y, & Z, 80% of the time, she's a pretty good poet by MY standards, but maybe only a 20% good poet by Po Land standards (assuming that A, B, & C are the elements comprising her other 20%).
This is an odd example, perhaps, but I hope it serves to show that any talk of "craft" is ultimately highly subjective, even if objective rules are in place.
About the Ellipticals--I was being flip, as I haven't read enough about Burt's classification to make an honest or insightful comment.
Finally, you are correct when you mention that I've left about many period styles--as you note, a complete period style taxonomy would be quite an undertaking.