Add `minimal_output` parameter to list_pull_requests and search_pull_requests

### Describe the feature or problem you'd like to solve

The `list_pull_requests` and `search_pull_requests` tools return extremely verbose responses that, even for small result
sets.

**Example:** Listing 30 PRs returned **508KB** of data (~17KB per PR), most of which is redundant metadata:
- Each PR includes full `user` objects with 15+ URL template fields
- `assignee` and `assignees[]` repeat the same user object structure
- `reactions` object with 10 fields (mostly zeros)
- Multiple `*_url` fields that follow predictable patterns

For a simple "show me my review queue" query returning 14 PRs, the response was ~36KB per PR.

### Proposed solution

Add `minimal_output` parameter (like `search_repositories` already has) to:
- `list_pull_requests`
- `search_pull_requests`

When `minimal_output: true`, return only essential fields:
```json
{
"number": 6617,
"title": "allow for no PDP filter in analytics",
"state": "open",
"draft": false,
"user_login": "ThomasRochais",
"created_at": "2026-02-02T13:05:54Z",
"updated_at": "2026-02-02T13:53:18Z",
"html_url": "https://github.com/org/repo/pull/6617",
"head_ref": "feature-branch",
"base_ref": "main"
}

This would reduce response size significantly.

Example prompts or workflows

1. "Show me PRs I need to review"
2. "List open PRs in this repo"
3. "Find PRs by author X"

Additional context

- search_repositories already implements minimal_output - this extends that pattern