Move more LOO functions into the array layer

[jordandeklerk]

Continuation of #267.

Migrates loo_score, loo_pit, loo_expectation, loo_quantile, loo_r2, and loo_summary into the base layer, making these functions accessible via .azstats without requiring the full LOO orchestration machinery.

Core functions require pre-computed PSIS weights rather than computing them internally. This enables weight reuse, e.g., users can compute PSIS once and pass the weights to multiple metrics, avoiding redundant computation. The exception to this is _loo, which can compute PSIS internally since it's the foundational function that produces pareto_k.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 96.86099% with 7 lines in your changes missing coverage. Please review.
✅ Project coverage is 84.11%. Comparing base (605327c) to head (928ce15).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/arviz_stats/base/dataarray.py	90.24%	4 Missing ⚠️
src/arviz_stats/base/diagnostics.py	97.27%	3 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #275      +/-   ##
==========================================
+ Coverage   83.88%   84.11%   +0.23%     
==========================================
  Files          42       42              
  Lines        5454     5560     +106     
==========================================
+ Hits         4575     4677     +102     
- Misses        879      883       +4     

☔ 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.

[jordandeklerk]

I think the RTD failure is unrelated to this

[aloctavodia]

I was not able to notice any obvious issue. In tests we trust!