The fact that you say, "They have their firm choice confirmed" is great news as it means you don't need to opt for the "
Priority review of marking" (request deadline 22 August) and can instead opt for the "Review of marking" (request deadline 26 September), which takes the time pressure off a little.
Correct. And the fact that there's really only one paper those marks can come from helps, and (a) it saves on cost (they charge per paper which is reviewed) and (b) you eliminate the risk of gaining marks when on paper is review only to lose them when the next paper is reviewed.
I can't tell you how many marks they'd have to lose to drop from a B to a C because you haven't mentioned the subject. You've just referred to it, someone elusively, as "a language". I could trying to match the weightings and maximum marks to the languages which AQA offer at A level, but it seems that you don't really want people to know what subject this was for some reason - so I'll respect that and not try to work it out.
The numbers I quoted in the post above came from Ofqual's
Reviews of marking and moderation for GCSE, AS and A level: summer 2022 exam series. Take care when reading it as they keep switching from whether they're talking about GCSE or A levels, 2019 or 2022, grade increases or mark increases. However Table 4 (in the main report) is probably most relevant to you ("Distribution of raw mark changes for AS and A levels following reviews") which shows that in 2022, 2.7% of reviewed papers went up by 4 marks and a further 5.5% went up by 5 marks or more. You will note, however, that 3.8% also went
down by 5 marks or more.
Again, without knowing the subject I can't confirm that it'd take 20 marks before dropping a grade. I can confirm, however, that the risk of dropping 20 marks is very, very low. But it's not zero.