attempt to find better starting values for GLMM when defaults aren't positive semidefinite

[palday]

closes #792

• add entry in NEWS.md
• after opening this PR, add a reference and run docs/NEWS-update.jl to update the cross-references.
• I've bumped the version appropriately

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 97.12%. Comparing base (7103ee1) to head (ffdf318).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #796      +/-   ##
==========================================
+ Coverage   97.08%   97.12%   +0.03%     
==========================================
  Files          33       33              
  Lines        3401     3405       +4     
==========================================
+ Hits         3302     3307       +5     
+ Misses         99       98       -1     

Flag	Coverage Δ
current	96.82% <100.00%> (+0.03%)	⬆️
minimum	97.06% <100.00%> (+0.03%)	⬆️
nightly	96.75% <100.00%> (+0.03%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

[dmbates]

On reading the section of src/generalizedlinearmixedmodel.jl that is modified it seems that a lot of work is being done to handle the case of constresponse. From the comments it seems that the reason for handling constresponse is for convenience in starting simulations. Is that correct? Would it be better to just error on constresponse with a message to come up with more realistic starting values?

[dmbates]

I see that I misread the code. You only try the fix-up if constresponse is false. Never mind.