fix: add checkHumanActor to agent mode

[ashwin-ant]

Fixes issue #641 where users were getting banned due to rapid successive Claude runs triggered by the synchronize event.

Changes:

• Add checkHumanActor call to agent mode's prepare() method to reject bot-triggered workflows unless explicitly allowed via allowed_bots
• Update checkHumanActor to accept GitHubContext (union type) instead of just ParsedGitHubContext
• Add tests for bot rejection/allowance in agent mode

[claude]

Consider adding a note in docs/custom-automations.md about this bot-checking behavior. Users who have automation workflows triggered by bots (like on synchronize events) will encounter the "non-human actor" error and need guidance on when/how to use allowed_bots.

Suggested section:

### Bot-Triggered Workflows

By default, automation mode rejects bot-triggered workflows to prevent infinite loops (e.g., Claude pushes code → `synchronize` event → Claude runs again). If you need to allow specific bots:

\`\`\`yaml
allowed_bots: "dependabot,renovate"  # Specific bots
# OR: allowed_bots: "*"              # All bots (use with caution)
\`\`\`

**Warning**: Be careful with `synchronize` events, as this can create loops if your workflow pushes commits.

[claude]

Consider adding two additional test cases:

1. GitHub API failure handling:

test("prepare method handles GitHub API failure gracefully", async () => {
  const mockOctokit = {
    rest: {
      users: {
        getByUsername: mock(() => Promise.reject(new Error("API Error"))),
      },
    },
  } as any;
  
  await expect(
    agentMode.prepare({
      context: contextWithPrompts,
      octokit: mockOctokit,
      githubToken: "test-token",
    })
  ).rejects.toThrow();
});

2. The actual bug scenario from issue #641:

test("prepare method prevents bot loop on synchronize/push event", async () => {
  const syncContext = createMockAutomationContext({
    eventName: "push", // The scenario that triggered issue #641
  });
  syncContext.actor = "claude-code[bot]";
  syncContext.inputs.allowedBots = "";
  
  const mockOctokit = {
    rest: {
      users: {
        getByUsername: mock(() =>
          Promise.resolve({
            data: { login: "claude-code[bot]", id: 12345, type: "Bot" },
          }),
        ),
      },
    },
  } as any;
  
  await expect(
    agentMode.prepare({
      context: syncContext,
      octokit: mockOctokit,
      githubToken: "test-token",
    })
  ).rejects.toThrow("Workflow initiated by non-human actor");
});

These tests would validate both error resilience and the specific scenario that caused the original issue.