fix(interferometer): resample non-PD inversion instead of crashing the search (#607)

AnalysisInterferometer.log_likelihood_function returned the fit figure_of_merit
unwrapped, so a non-positive-definite inversion matrix at a sampled model raised
a raw numpy.linalg.LinAlgError (Cholesky log-det, abstract.py:743) that
propagated through Nautilus and killed the whole search on the NumPy path
(release profile PYAUTO_DISABLE_JAX=1). The JAX path returns NaN and resamples,
masking it. Mirror the imaging analysis (imaging/model/analysis.py:132-144):
split on _use_jax, wrap the NumPy path in try/except -> af.exc.FitException,
which the fitness resamples. Apply the same guard to the point-source analysis,
which had the identical unguarded return.

Resolves release-validation tail item G (PyAutoHeart#72). Not jax-0.10.2 drift.

Co-authored-by: Jammy2211 <JNightingale2211@gmail.com>
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>

Author: 3 people

autolens/interferometer/model/analysis.py

@@ -167,10 +167,19 @@ def log_likelihood_function(self, instance, shared=None):
             instance=instance,
         )
 
-        return (
-            self.fit_from(instance=instance, preloads=shared).figure_of_merit
-            - log_likelihood_penalty
-        )
+        if self._use_jax:
+            return (
+                self.fit_from(instance=instance, preloads=shared).figure_of_merit
+                - log_likelihood_penalty
+            )
+
+        try:
+            return (
+                self.fit_from(instance=instance, preloads=shared).figure_of_merit
+                - log_likelihood_penalty
+            )
+        except Exception as e:
+            raise af.exc.FitException
 
     def shared_state_from(self, instance: af.ModelInstance):
         """

autolens/point/model/analysis.py

@@ -130,7 +130,13 @@ def log_likelihood_function(self, instance):
         float
             The log likelihood indicating how well this model instance fitted the imaging data.
         """
-        return self.fit_from(instance=instance).log_likelihood
+        if self._use_jax:
+            return self.fit_from(instance=instance).log_likelihood
+
+        try:
+            return self.fit_from(instance=instance).log_likelihood
+        except Exception as e:
+            raise af.exc.FitException
 
     def fit_from(
         self,