Adds test npy tensor for TuneAVideo pipeline tests for PR: #2455 diffusers

prompt = "A princess playing a guitar, modern disney style"
torch manual seed = 42
device = cuda
num_inference_steps = 50 ( tuneavideo-full-mo-di.npy ) ; = 2 (tuneavideo-2step-mo-di.npy)

text_to_video/tuneavideo-2step-mo-di.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:74c0696a43c35f1ee67e0af50795f18877ef3ffab87d64ed4fe3d0ec36b66a07
+size 9437312

text_to_video/tuneavideo-full-mo-di.npy

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9fd3e1b8fbab81e51e70b3e18be8ef97674fd7f0e7e505a0311864188a6322cc
+size 9437312