There is a lot to unpack here. I'll take it one at a time.
The auto-appeal of C-ratings is just based on the player's underlying year-end DNTRP. If it's within a certain tolerance, the appeal is granted. If it is not, the appeal is denied. If he was bumped, his DNTRP was 4.50 or higher. Since the appeal was granted, it wasn't higher than the appeal tolerance, so his year-end DNTRP was somewhere in the range of 4.50-4.55 (or whatever the exact tolerance is).
This is different than either medical appeals or self-rating appeals where the person has to fill out a more detailed form about either their background or their medical status and a committee of actual people look at it and decide.
For the other question, A-rated players can get strikes and get dynamically DQ'd. BUT, it is much more difficult to get DQ'd as an appeal than as a self-rate. The reason for this is that the player has a DNTRP that is barely over the bump threshold but significantly below the strike threshold, so the averaging of the match rating and existing DNTRP can be done as it is for C-rated players from the first match. For S-rated players, there is no existing DNTRP, so the match rating *IS* the DNTRP for the purpose of strike determination in their first two matches. Most S-rate DQ's happen because the self-rated player puts up a big match rating in their first match and can't recover from it when it starts to be averaged into the DNTRP. In other words, if the strike threshold for 4.5 is 4.70, then an A-rate with a starting DNTRP of 4.55 can play a match with a rating of 4.75 and the DNTRP is only 4.65 (average of 4.55 and 4.75). If he plays a second match at 4.75, it is also only a 4.65 DNTRP (average of 4.55, 4.65, and 4.75). Two matches over the strike threshold, no strikes. For the S-rate, the strike threshold is compared to the raw match rating (i.e. 4.75 in both cases) for the first two matches, and that player would have two strikes. In addition, if the player was bumped but only into the appeal range, then that player played at a 4.55 level the previous year. If that rating was not manipulated, then a sudden jump from playing 4.55 level tennis to 4.75 level tennis would be very unusual (i.e. not many people go from 4.5/5.0 borderline level to mid-5.0 or better between the end of one season and the start of the next). The easiest way to get strike level match ratings is to play up and win, so I'd recommend that he doesn't play in a 5.0 league, and the risk of DQ for an A-rate is very low.
As for singles vs doubles, I don't think that singles is more risky in terms of strikes unless he is very significantly better at singles. In other words, if he got a 4.55 rating as an average of playing 4.30 level doubles and 4.80 level singles, then he should probably save the singles for sectionals/nationals. But, if he's not some kind of singles hammer who is lost in doubles, then doubles might actually be more risky. The doubles calculation is more complex because there are 4 people to consider in the match ratings, but the partners in a doubles calculation maintain their starting rating differential in their individual match ratings. As an A-rate, your buddy will always be the higher rated partner, so if the rating differential is significant (i.e. he plays with a low-rated 4.5 partner), there is a danger of generating very high ratings because of the differential. In other words, if he is 4.55 and he plays with a 4.05 rated partner and they beat a top 4.5 team (so that their match rating as a team is 4.55 or something like that), your buddy would get a 4.80 match rating and his partner 4.30 to maintain the original 0.50 rating differential. If you're in a league where the higher rated players always play singles and you think you'll face lower rated players in doubles, then I'd recommend that he always has a higher-rated partner so that an anomaly with the rating differential doesn't generate strikes.