[Docs][Train] Replace Ray Tune + Train example with vanilla Ray Tune in homepage

[xinyuangui2]

Summary

Replaced the Ray Tune example in the homepage (index.html) to show vanilla Ray Tune usage instead of V1 tune+train integration.

Changes:

• Removed ScalingConfig and LightGBMTrainer imports (Ray Train components)
• Added a pure Ray Tune example demonstrating:
  • An objective function that trains a model with hyperparameters and reports metrics
  • Hyperparameter search space using common Tune methods (loguniform, choice, randint)
  • Running 1000 trials with the Tuner API
  • Retrieving the best result

This makes the example clearer for users who want to learn Ray Tune's hyperparameter optimization capabilities without the complexity of Ray Train integration.

[gemini-code-assist]

Code Review

This pull request effectively simplifies the Ray Tune example on the homepage by replacing a combined Ray Tune + Ray Train example with a vanilla Ray Tune one. This makes the core functionality of Ray Tune much clearer for new users. The new code snippet is well-structured, covering the definition of an objective function, a search space, and the execution of a tuning job. I have one suggestion to use tune.qrandint to avoid a common point of confusion with integer ranges, which will improve the example's clarity.

[matthewdeng]

thx