Fixed efficientnet v1 correct pad.

[sebastian-sz]

What does this PR do?

The correct_pad function in efficientnet v1 is using the wrong image_dim. It assumes channels_first, while keras-cv operates on channels_last.

Reference function: here

The weights should load anyway, but the metric values will be affected. (I'm not sure if this is big enough for a separate issue first)

Before submitting

• [x] This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• [ ] Did you read the contributor guideline, Pull Request section?
• [ ] Was this discussed/approved via a Github issue? Please add a link to it if that's the case.
• [ ] Did you write any new necessary tests?
• [ ] If this adds a new model, can you run a few training steps on TPU in Colab to ensure that no XLA incompatible OP are used?

Who can review?

@LukeWood