I know you shouldn't do this for a serious event...
On Sunday morning I had inherited the event computer for next week. Before uploading the results I did an estimate of times for the last two legs for those four competitors based upon comparison with times those who arrived before 2.30. I think the resulting final times have removed time spent looking for a non-existent control. A proper regression analysis might have been more accurate, but hardly justified. I did consider ending the course at the second last control, but then had a think about Livelox and decided I wasn't going there.
In defence of the collector of controls, the course setter was unable to be at the event so the Jones organised their second event in succession. I contributed to the difficulty by not being around to organise control collection due to a clash with a musical engagement. I think rebuilding event file was my penance.
Your revised results are a good estimate, although I would have been happy with the original results standing.