Add support for time-varying ascertainment

[seabbs]

As the title. On a single data source this could/would only be in the form of a user input but if any additional complexity (i.e. multiple data sources) it could be fit.

[seabbs]

An example use case could be if you know the number of hospitals that are reporting and this varies with time.

[kgostic]

Would you just specify the change point and let the model infer the scaling factor? Or would it be more like you have to input some proxy for the time-varying ascertainment across the entire time period?

[seabbs]

So I think you could do either and I think it would make sense to:

1. Most general case as a feature (i.e its all priros that vary over time)
2. Add a feature to make the fixed input special case
3. Add a helper to do piecewise changes

Starting with 1. makes sense as this then naturally extends to the planned model with multiple time series support

[sbfnk]

In principle this could be supported as part of the proposed changes in #525 and #600 if ever going down those routes.

[seabbs]

Yes agree. I was wondering about a easier version in the first instance though which could be supported right now with a special argument (for case 2.)

[sbfnk]

To be addressed following #547 with another column as discussed in #839.