combat LLM spam by building a web of trust

Tangled introduces a native vouching system to combat low-quality LLM-generated code submissions. Contributors receive green/red reputation shields visible to trusted circles, with maintainers able to vouch or denounce based on interaction history. Future iterations will add vouch decay and evidence trails linking vouches to specific pull requests.