This commit introduces the new `adk samples` command-line interface,
enabling users to easily discover, download, and prepare ADK example
agents for local use.

Key features in this initial version:

- Interactive Mode (`adk samples`):
  - Downloads official ADK samples from the GitHub repository to a local cache
    (default: `~/.adk/cache/samples/`, customizable with `--cache-path`).
  - Displays a table of available samples, including their names and extracted
    descriptions from their `agent.py` files.
  - Prompts the user to choose a sample by number.

- Direct Mode (`adk samples <sample_name>`):
  - Allows specifying a sample name directly for preparation.
  - Downloads samples to cache if not already present.

- Sample Preparation:
  - Copies the selected sample from the local cache to an `adk-samples`
    subdirectory in the user's target output path (default: current working
    directory, customizable with `--output-path`).
  - Prompts the user for their Gemini API key (securely).
  - Creates a `.env` file within the prepared sample's directory, pre-filled
    with the `GOOGLE_API_KEY` and configured for Google AI Studio backend.
  - Provides clear "Next steps" guidance on how to run the prepared sample
    using `adk web`, `adk run`, or `adk api_server`.

- Configuration:
  - Supports multiple sample sources via a configurable list (though V1
    defaults to the official adk-python repository).
  - Allows user-defined cache path (`--cache-path`) and output base path
    (`--output-path`).

File Changes:

- `src/google/adk/cli/cli_samples.py`: New file containing the core logic
  for downloading, caching, listing, copying samples, and .env creation.
- `src/google/adk/cli/cli_tools_click.py`: Modified to register the new
  `adk samples` command and its options (`--cache-path`, `--output-path`).
- `pyproject.toml`: Added `requests` as a dependency for downloading
  sample archives.