Use Hessian approximation in CyIpopt when grey-box without Hessian is present

[Robbybp]

Fixes #3891

Changes proposed in this PR:

• Set (potentially override) hessian_approximation=limited-memory in CyIpopt solver when the model contains a grey-box that doesn't support Hessians
• Log a warning if this happens and the user had set hessian_approximation=exact
• Add a has_hessian_support method to PyomoNLPWithGreyBoxBlocks (previously only existed on _ExternalGreyBoxAsNLP, although the former has a private _has_hessian_support attribute)

Legal Acknowledgement

By contributing to this software project, I have read the contribution guide and agree to the following terms and conditions for my contribution:

1. I agree my contributions are submitted under the BSD license.
2. I represent I am authorized to make the contributions and grant the license. If my employer has rights to intellectual property that includes these contributions, I represent that I have received permission to make contributions and grant the required license on behalf of that employer.

[codecov]

Codecov Report

❌ Patch coverage is 83.33333% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 89.93%. Comparing base (6baac63) to head (5ac3a48).
⚠️ Report is 54 commits behind head on main.

Files with missing lines	Patch %	Lines
...trib/pynumero/algorithms/solvers/cyipopt_solver.py	75.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3892      +/-   ##
==========================================
- Coverage   89.93%   89.93%   -0.01%     
==========================================
  Files         902      902              
  Lines      106393   106399       +6     
==========================================
+ Hits        95683    95685       +2     
- Misses      10710    10714       +4     

Flag	Coverage Δ
builders	29.21% <50.00%> (+0.01%)	⬆️
default	86.23% <83.33%> (?)
expensive	35.65% <50.00%> (?)
linux	87.38% <83.33%> (-2.04%)	⬇️
linux_other	87.38% <83.33%> (-0.01%)	⬇️
oldsolvers	28.12% <16.66%> (+0.01%)	⬆️
osx	82.73% <33.33%> (+0.01%)	⬆️
win	85.80% <83.33%> (-0.01%)	⬇️
win_other	85.80% <83.33%> (-0.01%)	⬇️

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.

🚀 New features to boost your workflow:

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

[jsiirola]

Looks good.

One design question, though: should we promote has_hessian_support up to PyomoNLP? (Maybe this should be added to a future "Let's revisit the design of Pynumero and ExternalGreyBox"?)

[jsiirola]

Can you add a test exercising / verifying this warning?

[Robbybp]

should we promote has_hessian_support up to PyomoNLP?

For now, no. PyomoNLP always has Hessian support. However, if we eventually merge grey box support into PyomoNLP (to eliminate the need for PyomoNLPWithGreyBoxBlocks), then we will need to do this.

[jsiirola]

should we promote has_hessian_support up to PyomoNLP?

For now, no. PyomoNLP always has Hessian support. However, if we eventually merge grey box support into PyomoNLP (to eliminate the need for PyomoNLPWithGreyBoxBlocks), then we will need to do this.

I was mostly thinking about an implementation like

def has_hessian_support(self):
    return True

and including it so that we had a "standard" interface API.

[Robbybp]

I see. I still would rather not add this before we need it.