[Bug]: ApiServer.__init__ has undefined type annotations EvalSetsManager and EvalSetResultsManager #7793

🔴 Required Information

Describe the Bug:
ApiServer.__init__ in src/google/adk/cli/api_server.py declares parameter type annotations eval_sets_manager: EvalSetsManager and eval_set_results_manager: EvalSetResultsManager, but neither class is imported in api_server.py (not even under if typing.TYPE_CHECKING:).

Because from __future__ import annotations is used, module import succeeds, but any static type checker (mypy) fails with [name-defined] errors:

src/google/adk/cli/api_server.py:1002: error: Name "EvalSetsManager" is not defined  [name-defined]
src/google/adk/cli/api_server.py:1003: error: Name "EvalSetResultsManager" is not defined  [name-defined]

Steps to Reproduce:

1. Run static analysis on api_server.py:

bash
uv run mypy src/google/adk/cli/api_server.py

2. Observe name-defined errors for EvalSetsManager and EvalSetResultsManager.

Expected Behavior: All type annotations used in ApiServer.init should be defined and resolve during static analysis.

Observed Behavior: Static analysis fails with undefined name errors for EvalSetsManager and EvalSetResultsManager.

Environment Details:

ADK Library Version: latest main
Python Version: 3.11+
OS: Windows / Linux / macOS
Model Information:

Are you using LiteLLM: No
Which model is being used: N/A

🟡 Optional Information

Regression: No

Additional Context:
Importing EvalSetResultsManager and EvalSetsManager under if typing.TYPE_CHECKING: resolves the missing type annotations for static analysis while maintaining lazy import behavior for runtime CLI performance (matching src/google/adk/cli/cli_tools_click.py).