Your approach is logical, using M15 for intraday trading will naturally generate more trade opportunities than the H2 swing trading examples shown in the guide, which makes sense for your prop firm challenge. Optimising every 2 months for the following month is also a reasonable cycle, as it keeps the parameters in line with recent market conditions while still providing some stability.
What you've noticed about the best optimisation result not consistently producing the best forward-month performance is common. Backtests are excellent for identifying promising parameter sets. Still, they can also “overfit” to the historical data, meaning that small market changes can cause the top result to perform worse out-of-sample. This is why many traders also test the 2nd- or 3rd-best optimisation sets; sometimes these slightly less “perfect” historical results can prove more stable in the future.
Suppose you're gathering data to compare which rank (1st, 2nd, 3rd, etc.) tends to produce the best forward results. In that case, you're essentially building your walk-forward optimisation process, which is a robust and well-respected method in systematic trading. Keep in mind that in rapidly changing markets, no single set will be optimal for long, so tracking and adapting over time is key.