FWIW, I became aware of the issue only after reading a
Eurocontrol document (on ARINC <-> AIXM mapping) that appeared to claim that TODA was TORA + stopway + (extra) clearway. That one-dimensional picture is not quite sufficient. OverRun's case E is a bit more explicit, in that it at least hints that CWY coould be shorter than SWY.
(For the record, I think the Eurocontrol team did an excellent job of AIXM.)