feat(prompt): add SiliconFlow LLM client support

[sky130]

Motivation

SiliconFlow (https://cloud.siliconflow.com/ or https://cloud.siliconflow.cn/) provides unified API access to multiple LLMs (DeepSeek, Qwen, GLM, etc.) with OpenAI-compatible endpoints. Adding this client allows Koog users to easily integrate these models using the same prompt executor patterns already established in the framework.

Changes

• Add prompt-executor-siliconflow-client module
• Implement SiliconFlowLLMClient (extends AbstractOpenAILLMClient)
• Include model definitions for supported SiliconFlow models
• Extend LLMModelParser with siliconflow provider identifier
• Add integration tests and module documentation

---

Note

The default endpoint is https://api.siliconflow.cn. To use https://api.siliconflow.com, configure the base URL manually.

[sky130]

Due to an accidental merge of the upstream branch, the commit history became cluttered with unrelated changes. After cleaning up with rebase, the local history diverged from the remote, requiring a force push to keep the PR history clean.