Compute bucket size at launch time

[jaredlockhart]

When we have real data for audience and features we'll be able to compute the number of buckets an experiment needs. Awaiting details from @tdsmith .

[jaredlockhart]

Blocked on data and feedback and design from @tdsmith

[tdsmith]

We can afford not to block on this for the A/A MVP; hard-coding 1% will give us the populations we need for that. This'll be more important when we gear up to run real experiments.

I think hashing out a spec for this in the MVP timeframe is reasonable, and I'll work on breaking this down, but the implementation can come later.

[jaredlockhart]

Blocked on mozilla/nimbus-shared#69

[jaredlockhart]

Blocked on https://github.com/mozilla/nimbus-shared/milestone/3

[jaredlockhart]

No fancy computing, it's just a constant:

https://github.com/mozilla/nimbus-shared/blob/main/data/ExperimentDesignPresets.ts#L37

[jaredlockhart]

What we need to do is create the actual bucket range object in the task here:

https://github.com/mozilla/experimenter/blob/main/app/experimenter/kinto/tasks.py#L23-L25

by calling this method:

https://github.com/mozilla/experimenter/blob/main/app/experimenter/experiments/models.py#L1330

and then updating the serializer to actually serialize the bucketrange object instead of returning a hard coded dict

and then THAT has to use the number of buckets from teh design template

[jaredlockhart]

Oh and the namespace is the normandy slug