refactor: eliminate test duplication via shared helpers

- Add tests/helpers.py with shared config, dataset, tokenizer, executor, and assertion helpers
- Replace 15 copies of device one-liner with session-scoped fixture
- Collapse 5 near-identical Dataset subclasses into RandomTokenDataset
- Remove duplicate _make_config/_make_model/_make_frozen and FakeTokenizer/FakeExecutor definitions
- Make test_callbacks and test_early_stopping use existing train_config_factory
- Replace 6 duplicate meta.json read blocks with load_shard_meta
- Fix mkdtemp leaks in test_lora.py with TemporaryDirectory

tests/trainer/test_early_stopping.py

@@ -1,43 +1,25 @@
 import os
 
-import numpy as np
-import torch
-
-from astrai.config.train_config import TrainConfig
-from astrai.trainer.schedule import SchedulerFactory
 from astrai.trainer.trainer import Trainer
+from tests.helpers import load_checkpoint_meta
 
 
-def test_early_stopping_simulation(base_test_env, early_stopping_dataset):
+def test_early_stopping_simulation(
+    base_test_env, early_stopping_dataset, train_config_factory, device
+):
     """Simulate early stopping behavior"""
-
-    def optimizer_fn(model):
-        return torch.optim.AdamW(model.parameters())
-
-    def scheduler_fn(optim):
-        return SchedulerFactory.create(
-            "cosine", optim, warmup_steps=10, lr_decay_steps=10, min_rate=0.05
-        )
-
-    train_config = TrainConfig(
-        strategy="seq",
-        optimizer_fn=optimizer_fn,
-        scheduler_fn=scheduler_fn,
+    train_config = train_config_factory(
         model_fn=lambda: base_test_env["model"],
         dataset=early_stopping_dataset,
-        ckpt_dir=base_test_env["test_dir"],
-        log_dir=os.path.join(base_test_env["test_dir"], "logs"),
+        test_dir=base_test_env["test_dir"],
+        device=device,
         n_epoch=2,
-        batch_per_device=2,
         ckpt_interval=1,
         grad_accum_steps=2,
-        random_seed=np.random.randint(1e4),
-        device_type=base_test_env["device"],
     )
 
     trainer = Trainer(train_config)
 
-    # Should handle early stopping gracefully
     try:
         trainer.train()
     except Exception:
@@ -50,8 +32,5 @@ def test_early_stopping_simulation(base_test_env, early_stopping_dataset):
 
     # Verify checkpoint was saved at expected step
     load_dir = os.path.join(base_test_env["test_dir"], "epoch_1_step_5")
-    import json
-
-    with open(os.path.join(load_dir, "meta.json")) as f:
-        meta = json.load(f)
+    meta = load_checkpoint_meta(load_dir)
     assert meta["consumed_samples"] == 20