[MLX LM] Sampler refactor + a few improvements (#1094)

* starting

* refactor sampler/processor and a few improvements

* fix stream

* fix stream generate

* fix eos handling in stream generate

llms/tests/test_generate.py

@@ -46,7 +46,7 @@ class TestGenerate(unittest.TestCase):
             "hello",
             max_tokens=5,
             verbose=False,
-            logits_processor=[logits_processor],
+            logits_processors=[logits_processor],
         )
         self.assertEqual(len(all_toks), len(init_toks) + 5)