The reason for this is that SC is built up from parallels. In the case of the Khandhakas, the Vinayas of all schools (with the partial exception of the Mahasanghika) include a Khandhaka section, which is divided into more or less the same set of chapters, as you can see from the parallels.
None of these Vinayas, however, divide their texts into a Mahavagga and Culavagga, apart from the Pali. So it seems that this division was a late development within the Pali schools, and probably represents a convenient way of dividing the manuscript.
Unlike the division into Khandhakas, the division as Mahavagga and Culavagga has no semantic basis: the text is simply a collection of 22 chapters (khandakas) arbitrarily divided in two.
For this reason we moved away from referencing in terms of Mahavagga and Culavagga, and reference simply to the khandhakas.
Even though we don’t use the names Mahavagga and Culavagga, no information has been lost, as you can still find the references to Vol 1 and Vol 2 of the PTS Pali edition, which correspond to the Mahavagga and Culavagga.
Generally speaking, this is a conceptual issue that underlies the structure of SuttaCentral. Most editions of the canon or canons work from the top down, starting with the canon, then Vinaya, Sutta and Abhidhamma, then the nikayas, and so on. SuttaCentral starts the other way around, from the sutta, and then builds the structures on top of that.
Since our aim is to get closer to the original form of the texts by organizing ourselves around the parallels between traditions, it is not possible to reproduce the details of referencing every canonical system. For example, in the Chinese texts as found on CBETA, an important referencing system is through the juan, or folio, which is a text-based division comparable to the division between Mahavagga and Culavagga. Since this system doesn’t exist in Pali, we can’t use it. But we always retain the metadata that enables referencing to be made.
In my opinion, this approach not only allows us to build sets of parallels, it is more useful since it focuses on the genuine semantic structures of the texts themselves. We shouldn’t remain bound to the pages of a particular printed edition, or to the arbitrary divisions that were once useful for manuscript copying. The texts had their own structure before those things ever existed, and that should be our primary means of reference.