fix: do not prefer float32 with tf experimental numpy behavior (#28928)

Author: Sam-Armstrong

ivy/functional/backends/tensorflow/__init__.py

@@ -18,7 +18,7 @@
 import ivy
 from ivy.func_wrapper import _dtype_from_version
 
-tf.experimental.numpy.experimental_enable_numpy_behavior(prefer_float32=True)
+tf.experimental.numpy.experimental_enable_numpy_behavior()
 
 backend_version = {"version": tf.__version__}