Write a function that produces intervention protocols for use by the ODE RHS

[ben18785]

See #20 for a description of how interventions are currently processed. This function may look something like:

Inputs:

• start date
• end date
• a coverage level

Outputs:

• A coverage vector \ matrix (where one column is time; the other coverage)

On the call, we discussed that probably the best way to handle this was by using "tanh" to smooth out the coverages so that they're continuous.

One thing to note is that, the way interventions are specified in the template means that the outputs should be superimposable. For example, you might have multiple school closures: (say) schools closed in march then opened April then closed again in June and opened up in July: in this case, I guess you'd want to be able to "sum" the two coverage "matrices" and make a single consistent matrix where coverage is bounded between 0 and 1. An alternative to handle this would be to take a vector of state and end dates and a vector of coverages.

[ben18785]

@DavidGavaghan posted Suli's notes on numerical solution of ordinary differential equations.

[ben18785]

I've also added @oscardm20994 since he's done this sort of thing before for interventions in Covid modelling.