Change success rate calculation

[manumoreira]

The current success rate calculation is not accurate, and impacts negatively in the way the survey broker (SB) works.
The typical use case is: a survey starts with Initial success rate at 0.5% and a target of 100. The SB calculates a huge Batch Size 200000. After one day of survey the survey gets 10 completed and 15 exhausted respondents (the 10 completes, plus 5 extra refused/failed/noneligible), then the Actual Success Rate is 66%, then the Estimated Success rate (ESR) is 6.7% (134 times bigger than the ISR) and consequently the batch size drops from 200000 to 1341. The SB already activated +10.000 respondents by now, which it won’t “deactivate”.