feat(billing): map gpt-image-2 quality (low/medium/high) to 1K/2K/4K tier

[yingcheng1026]

Summary

The billing layer already reads Group.ImagePrice1K/2K/4K via BillingService.CalculateImageCost, but normalizeOpenAIImageSizeTier classified strictly by pixel size, so the 4K price slot was unreachable for typical 1024x1024 requests even when callers asked for quality=high.

For gpt-image-2, image_tokens scale primarily with quality, not pixel dimensions:

$ curl … -d '{"size":"1024x1024","quality":"low"}'
  → output_tokens ≈ 196   (≈ 1K tier)

$ curl … -d '{"size":"1024x1024","quality":"high"}'
  → output_tokens ≈ 1756  (≈ 4K tier)

This PR makes normalizeOpenAIImageSizeTier accept (size, quality) and prefer quality-based classification. When quality is empty, "auto", or unrecognized it falls back to the existing size-dimension classification, preserving behaviour for older gpt-image-1 / dall-e flows and the OpenAI Responses API image_generation tool (image_generation_intent.go now also threads tool[\"quality\"] through).

• quality=low → 1K
• quality=medium / auto / empty → 2K (or size-based if size matches)
• quality=high → 4K

Why now

A user reported that requesting quality=high on gpt-image-2 (which costs 4× the tokens upstream) was billed at 1K tier in their group's ImagePrice1K, undercharging by 3×. Inspecting usage_logs confirmed image_size = "1K" for those rows, so the 4K column on groups was effectively dead code.

Test plan

• go build ./...
• go test -tags unit ./internal/service/... — all green
• Existing JSON parse test (size=1024x1024 + quality=high) updated to expect "4K" (the previous "1K" was the bug)
• Live deployment smoke test:
  • quality=low → usage_logs.image_size = "1K", total_cost = 0.20 (group ImagePrice1K = 0.2)
  • quality=high → usage_logs.image_size = "4K", total_cost = 0.60 (group ImagePrice4K = 0.6)

🤖 Generated with Claude Code

[github-actions]

Thank you for your contribution! Before we can merge this PR, we need you to sign our Contributor License Agreement (CLA).

To sign, please reply with the following comment:

I have read the CLA Document and I hereby sign the CLA

You only need to sign once — it will be valid for all your future contributions to this project.

---

I have read the CLA Document and I hereby sign the CLA

---

_{You can retrigger this bot by commenting recheck in this Pull Request. Posted by the CLA Assistant Lite bot.}