Add controlled emission HMM.

[rsenne]

Adds the ControlledHMM type we discussed here. This is a tested version of what I already proposed. I figured I would get things rolling and I can edit as we see fit. A few ongoing questions I had:

1. Should I add documentation for this new model? I'm happy to do so.
2. Do you want me document in the above or an existing tutorial making a custom emission model for a ControlledHMM? E.g., highlighting difference between the general HMM type like the extension of logdensityof?
3. Any additional tests/code documentation you would like?

Happy to work on any changes you'd like!

[codecov]

Codecov Report

❌ Patch coverage is 98.90110% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 95.83%. Comparing base (91f35ba) to head (48745c7).

Files with missing lines	Patch %	Lines
src/types/controlled_emission_hmm.jl	98.86%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #143      +/-   ##
==========================================
+ Coverage   95.28%   95.83%   +0.54%     
==========================================
  Files          18       19       +1     
  Lines         509      600      +91     
==========================================
+ Hits          485      575      +90     
- Misses         24       25       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[rsenne]

Okay so on the Standard test suite everything passes except the JuliaFormatter test which passes locally for me. Not sure why as I am using version 2.2.0 which seems to be what CI installs too. All tests related to the code (e.g., using the test_coherence) pass.

I also skipped the tests for the ControlledHMM when doing the HMMBase test suite. Not sure if this was the right thing to do.

[rsenne]

just wanted to ping this

[gdalle]

Sorry, I had quite a bit of time off work. Here are a few remarks, the most important one being on the semantics of obs_distributions and whether it is worth such a big change

[rsenne]

Okay so I implemented the ControlledEmission wrapper. I did have to re-add the log_transition_matrix(hmm::ControlledEmissionHMM, ::Nothing) = hmm.logtrans and transition_matrix(hmm::ControlledEmissionHMM, ::Nothing) = hmm.trans to resolve a method ambiguity. I'll take a look later today to see if there is a cleaner way around this, but the new API could be examined now

[gdalle]

Here are a few more remarks (I didn't go into detail yet while we refine the general structure)

[rsenne]

Hey @gdalle I got the error hint approach working. I also expanded the docs to reflect this new type--I left the old docs intact as they seem still helpful for insight into modifying the relevant HMM functions for custom types. Please let me know if there is anything else i can do

[gdalle]

Thanks! Can you try to hit the missing lines in codecov? The end after the @threads block is a false positive ofc

[rsenne]

All lines are now covered!