Add more magnitude DPFs

[sesquideus]

We should add more DPFs for magnitude discriminators, such as

• raised cosine CDF (https://en.wikipedia.org/wiki/Raised_cosine_distribution)
• arctangent
• bounded linear

Maybe a reverse approach and fitting would be appropriate too: find the number of seen and find a function by dividing by the number of simulated