Big TF test cleanup (#24282)

* Fix one BLIP arg not being optional, remove misspelled arg

* Remove the lxmert test overrides and just use the base test_saved_model_creation

* saved_model_creation fixes and re-enabling tests across the board

* Remove unnecessary skip

* Stop caching sinusoidal embeddings in speech_to_text

* Fix transfo_xl compilation

* Fix transfo_xl compilation

* Fix the conditionals in xglm

* Set the save spec only when building

* Clarify comment

* Move comment correctly

* Correct embeddings generation for speech2text

* Mark RAG generation tests as @slow

* Remove redundant else:

* Add comment to clarify the save_spec line in build()

* Fix size tests for XGLM at last!

* make fixup

* Remove one band_part operation

* Mark test_keras_fit as @slow

tests/models/bart/test_modeling_tf_bart.py

@@ -22,7 +22,7 @@ import unittest
 import numpy as np
 
 from transformers import BartConfig, BartTokenizer, is_tf_available
-from transformers.testing_utils import require_tf, slow, tooslow
+from transformers.testing_utils import require_tf, slow
 from transformers.utils import cached_property
 
 from ...test_configuration_common import ConfigTester
@@ -225,10 +225,6 @@ class TFBartModelTest(TFModelTesterMixin, TFCoreModelTesterMixin, PipelineTester
         config_and_inputs = self.model_tester.prepare_config_and_inputs_for_common()
         self.model_tester.check_decoder_model_past_large_inputs(*config_and_inputs)
 
-    @tooslow
-    def test_saved_model_creation(self):
-        pass
-
     # TODO (Joao): fix me
     @unittest.skip("Onnx compliancy broke with TF 2.10")
     def test_onnx_compliancy(self):