Fix Segment Anything Model saving bug

[tirthasheshpatel]

What does this PR do?

Loading keras_cv.models.SegmentAnythingModel presets fail when using Keras 3. This is because Keras 3 moves the random module outside of the ops namespace and one of the layers in the SAM model uses the ops.random.normal function. This PR fixes the error by accessing the function from the top-level keras module (keras.random.normal).

Before submitting

• 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?

Anyone in the community is free to review the PR once the tests have passed. Feel free to tag
members/contributors who may be interested in your PR.

@divyashreepathihalli @sampathweb

[divyashreepathihalli]

The keras 2 tests could be failing because of our recent changes. I will wait till we sort that out.

[tirthasheshpatel]

I think the Keras 2 failures are because of the changes here. Will push a commit to rosolve them. I am surprised the CI passes since, locally, a lot of them are failing because of is_tensor not being present in the Keras 3. Looks like the CI's been using Keras Core so those errors weren't caught! I think everything should be good once keras-team/keras#18756 is in.

[divyashreepathihalli]

/gcbrun

[divyashreepathihalli]

I think the Keras 2 failures are because of the changes here. Will push a commit to rosolve them. I am surprised the CI passes since, locally, a lot of them are failing because of is_tensor not being present in the Keras 3. Looks like the CI's been using Keras Core so those errors weren't caught! I think everything should be good once keras-team/keras#18756 is in.

the tests are not yet checking against keras 3. That is why.