Fixed segfault in Func module when using numpy

The C++ py_func_new function assumes it's being passed a 1-dimensional
array, but numpy generates 0-dimensional arrays for scalar inputs to
array / asarray, e.g. when using Func.Sin.

Cantera/python/Cantera/Func.py

@@ -43,7 +43,10 @@ class Func1:
         self._own = 1
         self._func_id = 0
         self._typ = typ
-        self.coeffs = asarray(coeffs,'d')
+        if _cantera.nummod == 'numpy':
+            self.coeffs = array(coeffs, dtype=float, ndmin=1)
+        else:
+            self.coeffs = asarray(coeffs,'d')
         self._func_id = _cantera.func_new(typ, n, self.coeffs)
 
     def __del__(self):