[SYSTEMDS-3259] Implementation of the shampoo optimizer (initial prototype)

[NicolasKorjahn]

Add Shampoo optimizer support for neural network training in SystemDS.
Includes full-matrix and diagonal preconditioning, momentum updates, and a heuristic variant with delayed preconditioner updates / infrequent root recomputation. Extend the existing NN training scripts with tests to validate correctness and convergence.

Tests: src/test/scripts/applications/nn/component/shampoo_test.dml
src/test/scripts/applications/nn/component/shampoo_test2.dml
src/test/scripts/applications/nn/component/shampoo_test.py

Experiments in staging/shampoo_optimizer

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 71.50%. Comparing base (b394e32) to head (1f54461).
⚠️ Report is 6 commits behind head on main.

Additional details and impacted files

@@             Coverage Diff              @@
##               main    #2401      +/-   ##
============================================
- Coverage     71.51%   71.50%   -0.01%     
+ Complexity    47441    47433       -8     
============================================
  Files          1539     1539              
  Lines        182605   182605              
  Branches      35916    35916              
============================================
- Hits         130585   130573      -12     
- Misses        42028    42032       +4     
- Partials       9992    10000       +8     

☔ 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.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[christinadionysio]

Thank you for the patch @NicolasKorjahn! Could you add the missing licenses and then I would merge it in. For your follow-up experiments etc. you can then create a new PR.