There's not really any spare units and the return trip would take 1:25 hours.
What I'm suggesting is that train A which is currently enters service at 08:25 does so at 07:43 to run a service to Okehampton, so that it's back in time to run the 09:25 service whilst the 08:25 service is run by a train which left at 06:43 which otherwise would have entered service at 07:25.
The 8:43 service from Waterloo then no longer needs to run the 9:25 service and so is free to run to Okehampton and back in time to run the 10:25 and so on through the day until the end of service, where the
As such, although there will be extra miles traveled for the units, there's no need for an extra unit.
Yes there would be extra costs (mostly staff costs and extra maintenance charges) but the actual lease of the units would be split. As such rather than a 2 coach train costing £250,000 it may well be that a 6 coach train on a 34/66 could cost GWR the same amount.
Although Okehampton may not have a large population (6,000), when added to the 7,600 of Crediton it could do rather well.
Part of the reason that Crediton has a low (~50,000) usage figure is down to the 08:05 or 09:47 arrival to Exeter St Davids of the morning peak hour services. Which is hardly useful for 9-5:30 job (although the evening is marginally better at 16:57 and 17:58 but still far from ideal given it's just 10 minutes on the train). Whilst a 08:25 arrival to and a 17:43 departure from Exeter St Davids would be significantly better for commuting.
Sorry, I'm not quite understanding this. If no additional units are used (which are unavailable, as your opening sentence says), which unit works the 07h25 departure (Exeter-London) if you've sent it to Okehampton at 06h43?