[generate] shape checks in tests compatible with fixed-length caches (+ some minor fixes) (#35993)

* shape checks compatible with static cache

* add test

* tmp

* manually turn on eager attn when we want to output attn

* typo

* generalize to encoder-decoder models

* force compilation on cpu

* tmp commit

* fix static cache shape checks

* models with odd caches

* fix copies

* shorter cache search loop

* use decoder_past_key_values everywhere

* better test variable names and comments

* signature

* rename _check_outputs into _check_generate_outputs

* add comments

* HybridCache future test note

tests/models/idefics/test_modeling_idefics.py

@@ -815,7 +815,7 @@ class IdeficsForVisionText2TextTest(IdeficsModelTest, GenerationTesterMixin, uni
                     )
 
     def _check_attentions_for_generate(
-        self, batch_size, attentions, min_length, max_length, config, use_cache=False, num_beam_groups=1
+        self, batch_size, attentions, prompt_length, output_length, config, decoder_past_key_values
     ):
         """
         Overwrite from generation tests because Idefics has only SDPA layers.