feat(diffusers): add AudioLDM2 generation

[localai-org-maint-bot]

Description

This PR adds AudioLDM2 text-to-audio generation through the existing diffusers backend and /v1/sound-generation API.

• implements the diffusers SoundGeneration RPC and writes pipeline output as mono PCM WAV
• maps duration_seconds to audio_length_in_s and prompt_influence to guidance_scale
• advertises sound generation in backend capability discovery
• adds the audioldm2 gallery model and user documentation
• adds focused request-mapping, WAV serialization, capability, and gallery coverage

Closes #1699

Notes for Reviewers

Verification:

• python3 -m py_compile backend/python/diffusers/audio_utils.py backend/python/diffusers/backend.py backend/python/diffusers/test.py
• focused PCM WAV serialization check
• go test ./core/config -ginkgo.focus="advertises diffusers sound generation"
• go test ./core/gallery -ginkgo.focus="gallery/index.yaml variant invariants|backend gallery entries are valid"
• git diff --check

The full go test ./core/config run passes 500 of 501 specs here. Its existing remote-config validation fails because this environment receives HTTP 403 from raw.githubusercontent.com.

The full diffusers runtime test needs the backend environment and model dependencies; this cron did not run a long backend build.

Signed commits

• Yes, I signed my commits.
• Documentation updated (docs/content/) for user-facing changes, or not applicable