Add LLM boundary hooks for request/response

[swk11rus]

What does this PR do?

Adds three plugin hooks at the LLM boundary: llm.request.before, llm.stream.chunk, and llm.response.after.
They run immediately before a provider request is sent, and immediately after a response chunk/result is received, enabling last‑mile masking and first‑mile unmasking without affecting opencode internal processing.

Changes are minimal and scoped to:

• plugin hook types
• LLM.stream (pre‑send + per‑chunk hook)
• Agent.generate (streamObject + generateObject hooks)

How did you verify your code works?

Manual test in my fork with a masking plugin:

• verified masking only happens right before request is sent
• verified unmasking happens before opencode processes streamed chunks
• verified non‑stream responses are unmasked at the boundary

[github-actions]

Hey! Your PR title Add LLM boundary hooks for request/response doesn't follow conventional commit format.

Please update it to start with one of:

• feat: or feat(scope): new feature
• fix: or fix(scope): bug fix
• docs: or docs(scope): documentation changes
• chore: or chore(scope): maintenance tasks
• refactor: or refactor(scope): code refactoring
• test: or test(scope): adding or updating tests

Where scope is the package name (e.g., app, desktop, opencode).

See CONTRIBUTING.md for details.

[github-actions]

The following comment was made by an LLM, it may be inaccurate:

No duplicate PRs found