fix: numpy-2.x safety in SamplesMCMC.median_pdf

np.percentile with a list [50] returns a shape-(1,) ndarray; float() on
that is a deprecated conversion that errors on numpy >= 2.4. Pass the
percentile as a scalar so the result is a scalar and float() is safe.

Unblocks searches/Emcee.py in autofit_workspace_test smoke tests on
Python 3.13 + numpy 2.4.

Co-Authored-By: Claude Opus 4.7 (1M context) <[email protected]>

Author: 2 people

autofit/non_linear/samples/mcmc.py

@@ -144,7 +144,7 @@ def median_pdf(self, as_instance: bool = True) -> [float]:
 
         if self.pdf_converged:
             return [
-                float(np.percentile(self.parameters_extract[i, :], [50]))
+                float(np.percentile(self.parameters_extract[i, :], 50))
                 for i in range(self.model.prior_count)
             ]