Add LoRA adapter support for video generation (#3306)

[Eshaan-byte]

This PR implements LoRA (Low-Rank Adaptation) adapter support for the LTX-Video transformer model, enabling users to apply community-created LoRA adapters to customize video generation. The implementation follows the same architecture pattern established for image generation LoRA support.

Changes Made:

Core Implementation:

• Removed blocking assertion that prevented adapter usage in ltx_video_transformer_3d_model.cpp
• Added adapter initialization logic in the compile() method
• Implemented set_adapters() method for runtime adapter control
• Added AdapterController member to manage adapter lifecycle
• Updated header file to include LoRA adapter dependencies

Sample & Tests:

• Created lora_text2video.cpp sample demonstrating:
  • Loading multiple LoRA adapters with custom alpha values
  • Generating videos with adapters enabled
  • Generating baseline videos for comparison
• Added comprehensive test suite in test_video_generation.py:
  • Test adapter support in pipeline constructor
  • Test adapter support in generate() method
  • Test transformer's set_adapters() method

User Benefits:

• Load and apply one or more LoRA adapters simultaneously
• Mix multiple adapters with different alpha (strength) values
• Enable/disable adapters per generation call
• Use community LoRA adapters from HuggingFace

Fixes #3306

Checklist:

• This PR follows GenAI Contributing guidelines.
• Tests have been updated or added to cover the new code.
  • Added TestLoRAVideoGeneration class with 3 test methods in tests/python_tests/test_video_generation.py
  • Tests cover: adapter constructor, adapter in generate(), and set_adapters() method
• This PR fully addresses the ticket.
  • All requirements from LoRA for Video Generation #3306 completed:
    ✓ Removed blocking assertion
    ✓ Wired up adapter support
    ✓ Added AdapterConfig support to Text2VideoPipeline
    ✓ Created sample demonstrating LoRA usage
    ✓ Added tests
• I have made corresponding changes to the documentation.
  • Documentation update to be added in follow-up (similar to image generation LoRA docs)

[likholat]

@Eshaan-byte Thanks for the contribution!
Some tests are currently failing on this PR. Could you please take a look?

[goyaladitya05]

Hi @Eshaan-byte, I've opened a PR against your fork (Eshaan-byte/pull/1) with the fixes for the CI failures.

The issue was a missing Python binding for set_adapters() and a missing null-check for the AdapterController. I'm currently working on #3313, which is blocked by this LoRA implementation, hence I wish to get this merged asap.

Once you merge my PR into your branch, the CI should pass. Please have a look.

[goyaladitya05]

@likholat, I've implemented the fixes for the CI failures here. Since I need this base support for #3313, I’m happy to open a consolidated PR (with eshaan's and my commits) to unblock this if @Eshaan-byte is unavailable to merge the fixes I sent to his fork.