feat: implement selective exports for modules 07-08

- 07_training: Export Trainer, CosineSchedule, clip_grad_norm only
- 08_dataloader: Export Dataset, DataLoader, TensorDataset only

Continues professional selective export pattern across all modules.
Development utilities remain in development, clean public API exported.

modules/source/07_training/training_dev.py

@@ -309,6 +309,7 @@ This preserves the relative magnitudes while preventing explosion.
 """
 
 # %% nbgrader={"grade": false, "grade_id": "gradient_clipping", "locked": false, "solution": true}
+#| export
 def clip_grad_norm(parameters: List, max_norm: float = 1.0) -> float:
     """
     Clip gradients by global norm to prevent exploding gradients.
@@ -485,6 +486,7 @@ for epoch in range(num_epochs):
 """
 
 # %% nbgrader={"grade": false, "grade_id": "trainer_class", "locked": false, "solution": true}
+#| export
 class Trainer:
     """
     Complete training orchestrator for neural networks.

modules/source/08_dataloader/dataloader_dev.py

@@ -44,14 +44,13 @@ Let's transform scattered data into organized learning batches!
 
 ## 📦 Where This Code Lives in the Final Package
 
-**Learning Side:** You work in modules/08_dataloader/dataloader_dev.py
-**Building Side:** Code exports to tinytorch.data.loader
+**Learning Side:** You work in `modules/08_dataloader/dataloader_dev.py`  
+**Building Side:** Code exports to `tinytorch.data.loader`
 
 ```python
-# Final package structure:
-from tinytorch.data.loader import Dataset, DataLoader, TensorDataset  # This module
-from tinytorch.data.loader import download_mnist, download_cifar10  # Dataset utilities
-from tinytorch.core.tensor import Tensor  # Foundation (Module 01)
+# How to use this module:
+from tinytorch.data.loader import Dataset, DataLoader, TensorDataset
+from tinytorch.data.loader import download_mnist, download_cifar10
 ```
 
 **Why this matters:**
@@ -127,6 +126,7 @@ Dataset Interface
 """
 
 # %% nbgrader={"grade": false, "grade_id": "dataset-implementation", "solution": true}
+#| export
 class Dataset(ABC):
     """
     Abstract base class for all datasets.
@@ -272,6 +272,7 @@ The key insight: TensorDataset transforms "arrays of data" into "a dataset that
 """
 
 # %% nbgrader={"grade": false, "grade_id": "tensordataset-implementation", "solution": true}
+#| export
 class TensorDataset(Dataset):
     """
     Dataset wrapping tensors for supervised learning.
@@ -467,6 +468,7 @@ This transforms the dataset from "access one sample" to "iterate through batches
 """
 
 # %% nbgrader={"grade": false, "grade_id": "dataloader-implementation", "solution": true}
+#| export
 class DataLoader:
     """
     Data loader with batching and shuffling support.