[codex] Fix inbound call log ingestion

[rahuliitk]

Summary

• Return organizationId, userId, agentNumber, and provider from the internal phone-number runtime config endpoint.
• Normalize call-log ingestion URLs so SERVER_API_URL=https://api.quickvoice.co posts to /api/v1/calls.
• Add a regression test for production-style call-log URL construction.

Root cause

Inbound calls were connecting, but completed call logs were not being saved. Production AI logs showed:

[CALL_LOG] Failed to post completed call: organization_id is required

The worker fetched config by phone number and received the agent configuration, but the server response did not include tenant metadata from the linked phone/agent. Without organizationId and userId, the AI worker could not build a valid call-log payload for the internal /calls ingest route.

There was also a second issue on the same path: post_call_log() posted directly to ${SERVER_API_URL}/calls. In production SERVER_API_URL is the origin (https://api.quickvoice.co), so the ingest URL needs /api/v1/calls.

Validation

• python3 apps/ai/tests/test_calllog_handler.py
• pnpm --filter server check-types
• pnpm --filter server build

Full AI unittest discovery is still blocked locally by missing optional runtime deps (loguru, livekit), unrelated to this focused call-log handler test.