Add hess_op for RegularizedNLPModels

[MaxenceGollier]

@dpo, in ExactPenalty.jl I defined L2PenalizedProblem <: AbstractRegularizedNLPModel
I will make an update RegularizedOptimization.jl, so that the solvers constructors call hess_op(reg_nlp, xk) at the start.
I will then dispatch a different hess_op(::L2PenalizedProblem) where I can give whatever I want.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 77.12%. Comparing base (85de184) to head (5001b10).
⚠️ Report is 17 commits behind head on main.

Additional details and impacted files

@@             Coverage Diff             @@
##             main      #81       +/-   ##
===========================================
- Coverage   99.18%   77.12%   -22.07%     
===========================================
  Files          11       19        +8     
  Lines         369      459       +90     
===========================================
- Hits          366      354       -12     
- Misses          3      105      +102     

☔ 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.

[dpo]

Those seem to be only required for the tests?!

[MaxenceGollier]

Yes, I added them in the [extras] section and not the [deps].
I am not sure however whether it is good practice or not to add packages from [extras] in the [compat] section... let me know.