fix(openai): report Responses prompt-cache tokens (TypeScript)

[didhd]

Description

The TypeScript OpenAIModel Responses adapter dropped OpenAI prompt-cache usage. mapResponsesEventToSDK built finalUsage from input_tokens / output_tokens / total_tokens only, so usage.input_tokens_details.cached_tokens was never surfaced as cacheReadInputTokens. Cache reads were therefore invisible in AgentResult usage metrics and OTel spans (gen_ai.usage.cache_read_input_tokens) for the Responses path, even when the API was returning cached_tokens.

This restores parity with the other adapters — models/bedrock.ts, models/anthropic.ts, and models/vercel.ts already plumb cacheReadInputTokens/cacheWriteInputTokens. The Usage type (models/streaming.ts) already declares both fields and the accumulator already sums them, so only the Responses adapter needed updating.

It is the TypeScript counterpart of the Python issue/PR (#2407 / #2663). The write side already works (a caller can pass prompt_cache_key / prompt_cache_retention via params, which formatResponsesRequest spreads into the request body); this PR makes the resulting cache reads observable.

Changes:

• Extract mapResponsesUsage() and call it from both the response.completed and response.incomplete branches (dedupes the existing mapping).
• Set cacheReadInputTokens only when cached_tokens > 0, matching the optional-field behavior of the other adapters.
• Add unit tests for a cache hit and the no-cache (cached_tokens: 0) case.

Related Issues

Fixes #2781

Documentation PR

No docs change needed — this aligns the Responses adapter with existing usage/telemetry behavior already documented for the other model providers.

Type of Change

Bug fix

Testing

strands-ts: npm run build (tsc, no errors), npm run lint (0 errors), npm test (full unit-node suite: 124 files / 3396 tests pass, no type errors). The two new tests in src/models/openai/__tests__/responses.test.ts cover the cache-hit and no-cache paths.

• I ran the TypeScript equivalents (npm run build, npm run lint, npm test)

Checklist

• I have read the CONTRIBUTING document
• I have reviewed and understand every line of code in this PR, including any generated by AI tools, and I can explain why it works
• My change is focused and reasonably small; I have split unrelated work into separate PRs
• I have added any necessary tests that prove my fix is effective or my feature works
• I have updated the documentation accordingly
• I have added an appropriate example to the documentation to outline the feature, or no new docs are needed
• My changes generate no new warnings
• Any dependent changes have been merged and published

---

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.