Ebola model based on Getz and Dougherty

[pratikunterwegs]

This PR adds an initial implementation of an SEIR ebola model using Erlang passage times, based on Getz and Dougherty (2017), and adpated from Epirecipes.

This PR also fixes #77, and this issue was only picked up during local testing on Ubuntu 23.04.

[pratikunterwegs]

Checks failing on Linux, and intermittently on Windows due to a memory error - will work this out in time.

[pratikunterwegs]

Calling @TimTaylor and @BlackEdder - any idea why this PR fails checks on Linux and Windows? I'm not able to fix or even diagnose the issue on my Linux machine.

[TimTaylor]

Errors for me. Fedora 38, gcc 13.1.1, R 4.3.1. Afraid I've no time to debug this so you probably need to spin up a Rocker Container and trying to reproduce in there.

[TimTaylor]

Errors for me. Fedora 38, gcc 13.1.1, R 4.3.1. Afraid I've no time to debug this so you probably need to spin up a Rocker Container and trying to reproduce in there.

Here is the error message though:

/usr/include/c++/13/bits/stl_vector.h:1125: std::vector<_Tp, _Alloc>::reference std::vector<_Tp, _Alloc>::operator[](size_type) [with _Tp = double; _Alloc = std::allocator<double>; reference = double&; size_type = long unsi
       │ gned int]: Assertion '__n < this->size()' failed.

Suggests that you've gone over a bound somewhere in the new ebola model.

[pratikunterwegs]

Thanks @TimTaylor, now debugging on my Ubuntu machine and get something different but related to bounds as well. Surprising that this passes on MacOS though.

[TimTaylor]

probably gcc being stricter than clang (in a good way). Maybe there are some flags you can set but 🤷

[pratikunterwegs]

Thanks - the issue turned out to be in .prob_discrete_erlang(). Still surprising since there are equivalence checks between this and the R implementation.