Without seeing an example application there isn't much that can be done. Since your example csv you provided are the "perfect" example of how not to receive such an error.
However, I will suggest to have a look at your joins and make sure they are unique as possible. What you are seeing will most likely be duplication of data on joins which is causing the calculation to sum more than you are actually expecting.
If you need further help, remove actual sensitive data from the application like company/client name address etc, also look into Settings->Document->Scrambling to scramble sensitive data to upload.
I would not expect the sum to not be right, but this is based on the sample you have provided. May be if I can guide you towards scrambling your data, you might be able to share you application? Try this link:
Yes the dimension names are different in real life (I have 2 CPM rates and a Flat Rate!)
This seems to work fine in a table and interestingly avoids all mention of summing - bizarrely I've tried to recreate the table and this expression returns nothing! But I have one working table at least.
What was basically going wrong when I was making my charts was just a placement of brackets.
For the first chart sum(Impressions) is absolutely fine and can be cut and sliced by any dimension.
When it comes to revenue it turn out you can't write, for example, sum(([Line Item Rate]*[DFP Impressions]/1000)) - it gives massively enlarged answers.
you can write
sum([DFP Impressions]/1000)*[Line Item Rate]
i.e. leave the summing to the impressions only
Furthermore to really get a chart that makes sense the aggr function needs to be thrown in