Implement the DRIFT model for delayed choice tasks

[drbenvincent]

• [x] implement the DRIFT model code
• [x] evaluate DRIFT model with a notebook
• [ ] add a parameter recovery simulation in the notebook
• [x] add reference to DRIFT paper in the code + notebook
• [ ] need to get a better sense of the parameterisation to get better priors over beta parameters

If all this works out:

• [ ] add it to the PsychoPy demos
• [ ] add it to tests