[BUG] Chapter 14: `DepthPool` class does not work when the number of channels is not divisible by the `pool_size`

In chapter 14, the following class is introduced:
```python
class DepthPool(tf.keras.layers.Layer):
def __init__(self, pool_size=2, **kwargs):
super().__init__(**kwargs)
self.pool_size = pool_size

def call(self, inputs):
shape = tf.shape(inputs) # shape[-1] is the number of channels
groups = shape[-1] // self.pool_size # number of channel groups
new_shape = tf.concat([shape[:-1], [groups, self.pool_size]], axis=0)
return tf.reduce_max(tf.reshape(inputs, new_shape), axis=-1)
```

The `tf.reshape` operation breaks down the last dimension to multiple groups so that then `tf.reduce_max` can be applied per group. However, the `tf.reshape` operation fails if `shape[-1]` (the number of channels) is not divisible by `self.pool_size`.

I suggest mentioning this limitation in the text. Alternatively, consider adding 3 lines to the `call` method to appropriately pad the tensor on the last dimension:
```python
class DepthPool(tf.keras.layers.Layer):
def __init__(self, pool_size=2, **kwargs):
super().__init__(**kwargs)
self.pool_size = pool_size

def call(self, inputs):
trailing_zeros = (self.pool_size - inputs.shape[-1] % self.pool_size) % self.pool_size
paddings = tf.constant([[0, 0], [0, 0], [0, 0], [0, trailing_zeros]])
inputs = tf.pad(inputs, paddings)

shape = tf.shape(inputs) # shape[-1] is the number of channels
groups = shape[-1] // self.pool_size # number of channel groups
new_shape = tf.concat([shape[:-1], [groups, self.pool_size]], axis=0)
return tf.reduce_max(tf.reshape(inputs, new_shape), axis=-1)
```

Open the Chapter 14 notebook and locate the DepthPool class shown in the issue. Check how it behaves when the channel count is not divisible by pool_size, then decide whether the chapter should document the limitation or support the case with padding. Done means the chosen behavior is explained or verified for non-divisible channel counts.

Written by the indexing model from the issue text.