[MRG] Debug sinkhorn divergence bug and add proper test (#394)

* skip tets if not torch installed

* update release.md

Author: rflamary

RELEASES.md

@@ -9,6 +9,7 @@
 
 #### Closed issues
 
+- Fixed an issue where sinkhorn divergence did not have a gradients (Issue #393, PR #394)
 - Fixed an issue where we could not ask TorchBackend to place a random tensor on GPU
   (Issue #371, PR #373)
 - Fixed an issue where Sinkhorn solver assumed a symmetric cost matrix (Issue #374, PR #375)

test/test_bregman.py

@@ -879,6 +879,7 @@ def test_empirical_sinkhorn_divergence(nx):
     ot.bregman.empirical_sinkhorn_divergence(X_sb, X_tb, 1, a=ab, b=bb, log=True)
 
 
+@pytest.mark.skipif(not torch, reason="No torch available")
 def test_empirical_sinkhorn_divergence_gradient():
     # Test sinkhorn divergence
     n = 10