feat(graphile-llm): add inference usage logging to metering plugin

[pyramation]

Summary

Adds inference usage logging to the LLM metering plugin. Every metered LLM call (embed or chat) now INSERTs into the usage_log_inference table with full usage metadata.

Changes:

• metering.ts: adds logInferenceUsage() helper that INSERTs into the inference log table; adds databaseId, actorId, inferenceLog to MeteringContext; both meteredEmbed and meteredChat now log usage on success and quota_exceeded
• metering-plugin.ts: extracts databaseId, actorId from pgSettings; queries inferenceLog config from getLlmBillingConfig()
• config-cache.ts: adds inferenceLog to the ModuleConfigCache — queries inference_log_module config alongside billing config
• Token counts use placeholderAmountTokens (~4 chars/token estimation) — clearly marked as placeholders pending team approval of generateWithUsage() on agentic-kit

Depends on: #1192 (billing metering + cache standardization, now merged)

Review & Testing Checklist for Human

• Verify usage_log_inference rows appear after an LLM embed/chat call with correct model, provider, request_type, token counts, latency_ms, and status
• Verify quota_exceeded events are logged (not just successes)
• Confirm placeholderAmountTokens variables are clearly placeholder — search for the comment "replace with actual provider token counts once generateWithUsage() is approved"

Notes

Token estimation is intentionally a placeholder (Math.ceil(text.length / 4)). The swap-out point is clearly marked in the code. Once generateWithUsage() is approved on agentic-kit, replace placeholderAmountTokens with actual provider counts.

Link to Devin session: https://app.devin.ai/sessions/2b5a29d83d3f478e8d3d972653b4879c
Requested by: @pyramation

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment and CI monitoring