refactor: deduplicate and restructure test suite

- extract preprocessing config factories into tests/data/factories.py
- keep conftest.py fixtures-only; stop importing builders from it
- promote temp_dir fixture to root conftest for cross-directory reuse
- unify duplicate BPE tokenizer builders into build_test_tokenizer
- merge grpo/dpo online e2e tests into one parametrized integration test
- extract engine mock factory and shared model batch builders
- drop local tempfile usage in favor of shared fixtures

No behavior change: 519 tests pass.

tests/conftest.py

@@ -5,12 +5,15 @@ import tempfile
 
 import pytest
 import torch
-from tokenizers import Tokenizer, models, pre_tokenizers, trainers
 
 from astrai.extension import KERNEL_NAMES, is_available
 from astrai.model.transformer import AutoRegressiveLM
-from astrai.tokenize import AutoTokenizer
-from tests.helpers import TINY_CONFIG, RandomTokenDataset, make_tiny_config
+from tests.helpers import (
+    TINY_CONFIG,
+    RandomTokenDataset,
+    build_test_tokenizer,
+    make_tiny_config,
+)
 
 CUDA_AVAIL = torch.cuda.is_available()
 KERNEL_AVAIL = CUDA_AVAIL and all(is_available(k) for k in KERNEL_NAMES)
@@ -30,18 +33,9 @@ def device():
     return "cuda" if torch.cuda.is_available() else "cpu"
 
 
-def create_test_tokenizer(vocab_size: int = 1000) -> AutoTokenizer:
+def create_test_tokenizer(vocab_size: int = 1000):
     """Create a simple tokenizer for testing purposes."""
-    tokenizer = Tokenizer(models.BPE())
-    tokenizer.pre_tokenizer = pre_tokenizers.ByteLevel()
-    trainer = trainers.BpeTrainer(
-        vocab_size=vocab_size, min_frequency=1, special_tokens=["<unk>", "<pad>"]
-    )
-    tokenizer.train_from_iterator([chr(i) for i in range(256)], trainer)
-    auto_tokenizer = AutoTokenizer()
-    auto_tokenizer._tokenizer = tokenizer
-    auto_tokenizer._special_token_map = {"unk_token": "<unk>", "pad_token": "<pad>"}
-    return auto_tokenizer
+    return build_test_tokenizer(vocab_size)
 
 
 @pytest.fixture(scope="session")
@@ -59,24 +53,29 @@ def test_model(device):
 
 
 @pytest.fixture
-def base_test_env(test_model, test_tokenizer):
+def temp_dir():
+    """Function-scoped temporary directory, cleaned up after each test."""
+    d = tempfile.mkdtemp()
+    yield d
+    shutil.rmtree(d, ignore_errors=True)
+
+
+@pytest.fixture
+def base_test_env(test_model, test_tokenizer, temp_dir):
     """Function-scoped test environment with isolated temp directory."""
-    test_dir = tempfile.mkdtemp()
-    config_path = os.path.join(test_dir, "config.json")
+    config_path = os.path.join(temp_dir, "config.json")
     with open(config_path, "w") as f:
         json.dump(TINY_CONFIG, f)
 
-    yield {
+    return {
         "device": test_model["device"],
-        "test_dir": str(test_dir),
+        "test_dir": temp_dir,
         "config_path": config_path,
         "transformer_config": test_model["config"],
         "model": test_model["model"],
         "tokenizer": test_tokenizer,
     }
 
-    shutil.rmtree(test_dir)
-
 
 @pytest.fixture
 def random_dataset():