Oh, but subject/complement order IS one of those points that determines whether a clause is canonical or non-canonical.
If you look at SIEG page 24, you’ll see that it says:
"Canonical clauses consist of a subject followed by a predicate ...".
But the reversal of subject and PC means that the predicate is followed by a subject so the clause is not canonical. Compare the two examples where the subject is "the next point":
The next point is more serious. (subject followed by predicate = canonical)
More serious is the next point. (predicate followed by subject = non-canonical)
You must remember that canonical clauses package the information in the grammatically most basic way, which means the subject must be at the front, not at the end.
I was just comparing it with the benchmark definition while thinking the order of information packaging is a bit inferior. But it is indeed to be considered.
All the constructions considered to be 'information packaging' (see Ch15) are non-canonical.
The reversal of subject and PC falls under the heading of 'information packaging', subtype 'postposing and preposing', more specifically subject-dependent inversion (see Sect 8, page 258 [47] iia/iib).
Examples of the five major canonical clause structures are given in Ch4, p78.