Remove workaround for obsolete TF bug.

PiperOrigin-RevId: 380254090

Author: brianwa84

tensorflow_probability/python/experimental/mcmc/elliptical_slice_sampler_test.py

@@ -160,9 +160,6 @@ def normal_log_likelihood(state):
         seed=test_util.test_seed(),
         trace_fn=None)
 
-    # TODO(b/159923910): Retval[0] does not have a value.
-    samples = tf.convert_to_tensor(self.evaluate(samples))
-
     mean, variance = self.evaluate(tf.nn.moments(samples, axes=[0]))
     posterior_mean, posterior_variance = normal_normal_posterior(
         prior_mean=0.,