docs(rho_setting): document grad-rho options + flag sep-rho/reduced-costs-rho deprecations

[DLWoodruff]

Summary

Doc-only follow-up to #679. Two changes:

1. Document the gradient-rho options. rho_setting.rst previously
   described only --grad-rho-multiplier out of the six options
   cfg.gradient_args() registers. Add a bullet list covering all
   five undocumented options:

   • --grad-order-stat (default 0.5) — order statistic across
     scenarios; 0 = min, 1 = max, 0.5 = mean, with linear interpolation
     in the open intervals.
   • --grad-rho-relative-bound (default 100) — floor on the
     primal-residual denominator relative to xbar.
   • --eval-at-xhat (default off) — evaluate the gradient at the
     best xhat from an xhat-producing spoke instead of at the latest
     subproblem iterate.
   • --indep-denom (default off) — use a single
     probability-weighted, MPI-Allreduced denominator across scenarios.
   • --grad-rho-multiplier (default 1.0) — scalar multiplier
     applied to every computed rho (already documented; left in for
     completeness in the same list).

2. Flag --sep-rho and --reduced-costs-rho as scheduled for
   deprecation. Per
   issue #673 <https://github.com/Pyomo/mpi-sppy/issues/673>_, both
   methods are slated for removal: SepRho is expected to be
   subsumed into GradRho, and reduced-cost rho has not been
   demonstrated to be effective in practice. PR rho setters: aggregate per-sub-scenario nonants in proper bundles (closes #673) #679 has already
   added a DeprecationWarning at instantiation; this PR adds an
   .. note:: on each section so the rendered docs match.

grad_rho.rst stays a brief pointer (the academic paper covering the
algorithm is in preparation); a .. _grad_rho: cross-reference label
is added so rho_setting.rst can link to it.

Test plan

• cd doc && make html builds without new warnings (verified
  locally — the only warning is the pre-existing
  viewcode_import config typo, unrelated).
• Read the Docs preview renders the new option list and deprecation
  notes correctly.
• No code changes; non-docs CI is N/A.

🤖 Generated with Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 71.03%. Comparing base (bfc2ce3) to head (9f0c54d).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #680      +/-   ##
==========================================
+ Coverage   71.02%   71.03%   +0.01%     
==========================================
  Files         154      154              
  Lines       19248    19248              
==========================================
+ Hits        13670    13672       +2     
+ Misses       5578     5576       -2     

☔ 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.