llm is the only app (#15779)

* tinygrad/llm is the only app

* upd pyproject

* claude refs

* scoping

* min diff

test/null/test_attention.py

@@ -1,6 +1,6 @@
 import unittest
 from tinygrad import Tensor, dtypes, TinyJit, UOp
-from tinygrad.apps.llm import apply_rope as apply_rope_new, precompute_freqs_cis
+from tinygrad.llm.cli import apply_rope as apply_rope_new, precompute_freqs_cis
 from test.helpers import assert_jit_cache_len
 
 def apply_rope(x:Tensor, start_pos:int):

test/null/test_llm_server.py

@@ -22,18 +22,15 @@ class TestLLMServer(unittest.TestCase):
     cls.bos_id = 1
     cls.eos_id = 999
 
-    import tinygrad.apps.llm as llm_module
-    llm_module.model = cls.mock_model
-    llm_module.model_name = "test-model"
-    llm_module.tok = cls.mock_tok
-    llm_module.bos_id = cls.bos_id
-    llm_module.eos_id = cls.eos_id
-    llm_module.eot_id = None
+    from tinygrad.llm.cli import Handler, LLMServer
 
-    from tinygrad.apps.llm import Handler
-    from tinygrad.viz.serve import TCPServerWithReuse
-
-    cls.server = TCPServerWithReuse(('127.0.0.1', 0), Handler)
+    cls.server = LLMServer(('127.0.0.1', 0), Handler)
+    cls.server.model = cls.mock_model
+    cls.server.model_name = "test-model"
+    cls.server.tok = cls.mock_tok
+    cls.server.bos_id = cls.bos_id
+    cls.server.eos_id = cls.eos_id
+    cls.server.eot_id = None
     cls.port = cls.server.server_address[1]
     cls.server_thread = threading.Thread(target=cls.server.serve_forever, daemon=True)
     cls.server_thread.start()

test/null/test_llm_tokenizer.py

@@ -1,5 +1,5 @@
 import unittest, base64, functools, sys
-from tinygrad.apps.llm import SimpleTokenizer
+from tinygrad.llm.cli import SimpleTokenizer
 from tinygrad.helpers import fetch
 
 @unittest.skipIf(sys.platform == 'win32', "fetch race condition on Windows")