Add images api

[jglee-sqbits]

Description

This PR introduces the Diffusion pipeline along with the serving infrastructure. It covers the entrypoint setup and image request handling.

Testing:

• Run the command below to verify the changes:
• Run with pipelines_diffusion for torch dependency(cudnn fallback)

./bazelw run //max/python/max/entrypoints:pipelines_diffusion -- \
  images generate \
  --output /absolute/output/image/path \
  --prompt "a cat in the jungle" \
  --model black-forest-labs/FLUX.1-dev \
  --seed 4242

./bazelw run //max/python/max/entrypoints:pipelines_diffusion -- serve --model black-forest-labs/FLUX.1-dev --task image_generation

curl test

python max/examples/diffusion/client_example.py 

openai api test

python max/examples/diffusion/openai_api_example.py 

Key Changes:

• Diffusion Pipeline: Added the diffusion pipeline with queue integration.

• Serve: Implemented the serve logic.

• Entrypoint: Added the application entrypoint.

• Image Processing:

  • Added Image Request handling.

  • Added Image Pipeline.

⚠️ Important Notes:

• WIP (Work In Progress): This PR will be maintained and updated until the serve implementation is fully complete. It will be merged afterwards.

• Examples: Please note that the provided examples might not work as expected in this current state.