I reckon it is a case of BT being pragmatic, and working out the lowest cost vs the impact ie how many people end up within SF range of the FTTC cabinet, compared with the number that don't.
As you say, adding a cabinet outside the exchange is a solution that works for EO lines within range of the exchange buildings, but doesn't work if the lines are long.
Re-parenting is possible when EO lines happen to route close to an existing PCP cabinet that has capacity to terminate the lines (or can be increased in size cheaply), but obviously only works well if the EO premises are then close enough to the cabinet. I have heard of this technique before, fairly recently, but I don't think we've seen any concrete examples of it taking places yet.
A third option is to add a new cabinet for the long EO lines, sited a long way from the exchange building, but close to the cluster of buildings. That of course requires a reasonable cluster to exist, sized to work well with one of the cabinet sizes used by Openreach.
We know North Yorkshire is due to trial a new form of cabinet that BT term "remote node" (hence FTTRN); we don't know any details about how they intend to deploy this, but my suspicion is that it will be used in this 3rd way.