I think I can explain this, but my diagnosis may be out of date since Sage now uses more pari functions than it used to, and fewer of the function I wrote for Sage itself.
The point is a point on an elliptic curve E over the field GF(103^20) but the curve's j-invariant only has degree 2, which means that computing the cardinality of E is fast, the result being
sage: E.cardinality()
18061112346694138117304349799948434434304
sage: E.cardinality().factor()
2^8 * 3^2 * 11^2 * 13^2 * 17^2 * 1171^2 * 95191^2 * 10332211^2
Knowing the cardinality of E makes finding the order of points straightforward. Now asking a point for its order might check whether the curve cardinality is known (I assume that this is true -- a pari function is called), but does not trigger its calculation if not. Now when you ask Sage for the cardinality it *does* check whether the j-invariant lies in a subfield, but the cardinality function in pari, which is called implicitly when you ask for the order of P first, does not (as I have been told by pari developers).
It might be a good idea for Sage to do something different and possibly compute the cardinality of E before calling pari for the order of the point. One reason why we used not compute the curve cardinality first always was that the methods of computing the cardinality used to (and perhaps still do) involve computing the order of random points.
I myself do not intend to go further with this. Perhaps the people who replaced my code with calls to pari would like to comment?
John