AI-as-Infrastructure/aiinfra-atlas: Feedback inter-rating

This release adds functionality for inter-rating of user feedback. Refer to /docs/inter_rater.md for configuration details.

Major Features:

• Numbered inter-rater annotations ([inter-rating-1], [inter-rating-2], [inter-rating-3])
• Prevents annotation overwrites with unique IDs per rater
• Async event loop compatibility for concurrent users
• Enhanced documentation with allocation algorithm details

Key Changes:

• Made submit_span_annotation() async for proper FastAPI integration
• Include inter_rater_number in annotation IDs and metadata
• Use async httpx.AsyncClient for non-blocking Phoenix queries
• Deterministic user-specific session allocation

Fixes:

• Fixed inter_rater_number always being null (GitHub issue #43)
• Fixed annotation overwrites when multiple raters rate same session
• Fixed async/await event loop conflicts in FastAPI

Documentation:

• Inter-rating configuration guidelines for different team sizes
• Coverage calculation formulas and examples
• Detailed allocation algorithm explanation
• Workload estimation guidance

Related: GitHub issue #43