Do not use initval in test model

[ricardoV94]

PRs #7508 and #7492 introduced incompatible changes but were not tested simultaneously.

Deepcopying the steps in the tests leads to deepcopying the model which uses clone_model, which in turn does not support initvals.

See #7528 why this is currently not implemented.

@lucianopaz irregardless of the issues with copying models with initvals, is there any concern for the step functionality that model.copy uses clone_model these days? AFAICT this was only used in the tests so a first pass I would guess no?

---

📚 Documentation preview 📚: https://pymc--7529.org.readthedocs.build/en/7529/

[lucianopaz]

@lucianopaz irregardless of the issues with copying models with initvals, is there any concern for the step functionality that model.copy, uses clone_model these days? AFAICT this was only used in the tests so a first pass I would guess no?

No, there should be no issue with that. The step methods sampling_state is independent from the model instance and its variables, so as long as copying the model doesn't break the relationship with the step method's var, then nothing should break.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.68%. Comparing base (465d8ac) to head (79200d2).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #7529      +/-   ##
==========================================
+ Coverage   90.55%   92.68%   +2.13%     
==========================================
  Files         104      104              
  Lines       17412    17412              
==========================================
+ Hits        15767    16139     +372     
+ Misses       1645     1273     -372     

see 16 files with indirect coverage changes