Re-implement token caching for Vercel AI SDK usage.

[bhouston]

We used to have token caching implemented for the Anthropic SDK when using it directly but now that we are moving towards the Vercel AI SDK, we need to re-implement token caching using the Vercel AI SDK method. The way to do it is specific per AI provider, so the Anthropic way is described here: https://sdk.vercel.ai/providers/ai-sdk-providers/anthropic

[bhouston]

There is commented out code in toolAgent.ts which added the proper cache_control data for the direct Anthropic integration. This would have to be redone for the Vercel AI SDK integration, but it should be quite similar.

[github-actions]

🎉 This issue has been resolved in version mycoder-agent-v1.0.0 🎉

The release is available on:

• npm package (@latest dist-tag)
• GitHub release

Your semantic-release bot 📦🚀