Handle incompatible issues with numba and c unboxing of boolean scalars

Author: ricardoV94

pytensor/tensor/signal/conv.py

@@ -255,8 +255,11 @@ def __init__(self, method: Literal["direct", "fft", "auto"] = "auto"):
     def perform(self, node, inputs, outputs):
         in1, in2, full_mode = inputs
 
-        # TODO: Why is .item() needed?
-        mode: Literal["full", "valid", "same"] = "full" if full_mode.item() else "valid"
+        if isinstance(full_mode, np.bool):
+            # Patch for wrong unboxing of bool scalars in C backend
+            # Conditional, because numba will produce a bool, not np.bool_
+            full_mode = full_mode.item()
+        mode = "full" if full_mode else "valid"
         outputs[0][0] = scipy_convolve(in1, in2, mode=mode, method=self.method)