stan model implementation of simulate_secondary

[sbfnk]

simulate_secondary is a nice function for convolutions with potentially time-varying delays and scaling, implemented in R.

Whilst this is great I think it would be nice to have a simulate_secondary function that mirrors simulate_infections and uses the corresponding stan model to simulate from given parameters. This could benefit from further model updates (and already e.g. can handle gamma distributed delays) and e.g. the new distribution interface but for the moment wouldn't be able to have time-varying convolution parameters. Implementing this would also enable the deprecations mentioned in #520.

We could

1. replace simulate_secondary with the function outlined above
2. keep simulate_secondary and add the function outlined above with different name
3. rename simulate_secondary and add the function outlined above as simulate_secondary
4. not do anything and keep things as they are

My initial inclination would be (3) with an aim to eventually deprecate the function when time-varying convolutions become available in the stan model.

[seabbs]

agree

[sbfnk]

Moving this to 1.5.0 so as to be able to bring forward #520 in the aim for a "cleaner" 2.0.0.