HomogeneousPoissonProcess type

[JoseKling]

Although it would be possible to use either MarkedPoissonProcess or MultivariatePoissonProcess to emulate a homogeneous Poisson process, it would be a bit awkward and very unintuitive. Since this is the basic process, I guess this should have its own type and be convenient to use.

Having this will also make it easier to implement tests for the goodness-of-fit tests in PR #60.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[jucheval]

If you chose this implementation, I guess that it would make sense to close PR #51: its objective was to consider MultivariatePoissonProcess (and in particular univariate ones) as a particular case of MarkedPoissonProcess.

[JoseKling]

Ah, true. I saw it some time ago, but forgot about it. I guess merging everything into one single type makes sense. I will have a better look at it.

Why wasn't it merged? What is still missing?

[jucheval]

There is still a problem with automatic conversion made by Distributions.jl. If I am correct, it only concerns Multivariate Poisson processes and not Univariate ones. As suggested by @gdalle, we could just decide that it is not our fault and recommend users to use only Float intensities and not Integer ones.

[JoseKling]

Just left a comment there.

I played around a while and I think this is quite an elegant solution. And it can be extended to other processes as well, like the HawkesProcess.

I will close this PR, then.