Upgrades all around

[gdalle]

Breaking

• Remove AbstractModel to make AbstractHiddenMarkovModel a subtype of AbstractMarkovChain

Features

• New function forward
• (Not working) chain rule for reverse mode autodiff of `logdensityof(hmm, obs_seq)

Improvements

• Better docstrings
• Precompile on parallel versions of inference algos to hit more code paths

[codecov]

Codecov Report

Patch coverage: 85.00% and project coverage change: -2.18% ⚠️

Comparison is base (bca1bab) 93.16% compared to head (b2050fb) 90.99%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #32      +/-   ##
==========================================
- Coverage   93.16%   90.99%   -2.18%     
==========================================
  Files          19       18       -1     
  Lines         483      522      +39     
==========================================
+ Hits          450      475      +25     
- Misses         33       47      +14     

Files Changed	Coverage Δ
src/inference/viterbi.jl	95.45% <ø> (ø)
src/types/abstract_hmm.jl	90.90% <ø> (ø)
ext/HiddenMarkovModelsChainRulesCoreExt.jl	13.33% <13.33%> (ø)
src/inference/forward.jl	92.15% <88.23%> (ø)
src/HiddenMarkovModels.jl	100.00% <100.00%> (ø)
src/inference/baum_welch.jl	90.69% <100.00%> (ø)
src/inference/forward_backward.jl	97.11% <100.00%> (+1.52%)	⬆️
src/types/abstract_mc.jl	97.29% <100.00%> (+0.07%)	⬆️

... and 1 file with indirect coverage changes

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