The appended member of a divisor family is its last member #
In the maximal contact case of the proof of Theorem 103, Kollár declares E^0 := H to be the
first divisor of H + E and applies Lemma 102 with j = 0 [Kol07, 104, Step 2.2]. In this
library DivisorFamily.append E J puts the new member last (its index set is E.ι ⊕ₗ PUnit,
every inl before the inr), and DivisorFamily.nth counts positions from 0 in the linear
order of the index set. So Kollár's position 0 is the position card E.ι here:
(E.append J).nth ⟨card E.ι, _⟩ = J (nth_append_last). The finite-order fact behind it is that
the last position of a lexicographic sum with a one-element right summand is that summand
(monoEquivOfFin_lex_inr_last), the counterpart for inr of monoEquivOfFin_lex_inl.
The last position of a lexicographic sum of finite linear orders whose right summand has a top
is the top: an order isomorphism Fin k ≃o α ⊕ₗ β sends the last index to the greatest element.
card (α ⊕ₗ PUnit) = card α + 1.
In α ⊕ₗ PUnit the element at position card α, the last one, is the inr; the counterpart
for inr of monoEquivOfFin_lex_inl.
One member is appended: card E.ι < card (E.append J).ι.
The appended member J of E.append J sits at the last position card E.ι (positions are
counted from 0, new members come last). Kollár's instruction to "declare E^0 := H to be the
first divisor in H + E" and apply Lemma 102 with j = 0 [Kol07, 104, Step 2.2] therefore reads
j := card E.ι here.