Stabilize gp.Latent.conditional covariance with jitter (fixes #8108)

[mullachv]

Stabilize gp.Latent.conditional covariance with jitter (fixes #8108)

Description

With large Xnew arrays, gp.Latent.conditional could raise LinAlgError: Matrix is not positive definite because the conditional covariance matrix was not numerically stabilized. The jitter argument was only applied to the prior covariance K(X,X) before Cholesky, not to the conditional covariance K(Xnew,Xnew) - K(Xnew,X) K(X,X)^{-1} K(X,Xnew).

Changes:

• Apply stabilize(..., jitter) to the conditional covariance in Latent._build_conditional so jitter is used for conditional covariance passed to MvNormal.
• Same in TP._build_conditional for consistency.
• Add a regression test testLatentConditionalLargeXnewJitter that builds the model from issue BUG: gp.Latent.conditional conditional cov needs stabilizing? #8108 (99 training points, 500 prediction points, ExpQuad, jitter=1e-4) and evaluates the compiled logp to ensure no LinAlgError.

Related Issue

• Closes BUG: gp.Latent.conditional conditional cov needs stabilizing? #8108
• Related to #

Checklist

• Checked that the pre-commit linting/style checks pass
• Included tests that prove the fix is effective or that the new feature works
• Added necessary documentation (docstrings and/or example notebooks)
• If you are a pro: each commit corresponds to a [relevant logical change]

Type of change

• New feature / enhancement
• Bug fix
• Documentation
• Maintenance
• Other (please specify):

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 91.49%. Comparing base (c32715a) to head (6f81f0c).
⚠️ Report is 3 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #8109   +/-   ##
=======================================
  Coverage   91.49%   91.49%           
=======================================
  Files         123      123           
  Lines       19832    19832           
=======================================
  Hits        18146    18146           
  Misses       1686     1686           

Files with missing lines	Coverage Δ
pymc/gp/gp.py	94.71% <100.00%> (ø)

🚀 New features to boost your workflow:

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