In a very broad sense, there are two "perspectives" you can take when dealing with sampling.
1. You know the population distribution (ie. the proportion of individuals in the entire population with a given characteristic), and you want to calculate the probability that a random sample will have a given proportion of individuals with that characteristic. These are the questions where you are asked to find the mean / standard deviation of the distribution of sampling proportions. For these, use the proportion within the population - in this case, 0.9.
2. You know the proportion of individuals with a given characteristic in a random sample, and you want use that sample to infer a reasonable estimate of the proportion of individuals with that characteristic in the entire population. These are the questions where you are asked to find a confidence interval for the population proportion. For these, use the proportion within the sample - in this case 47/51.