Add Gemini CLI provider support and fix output parsing issues

[stn1slv]

This pull request adds support for the Gemini CLI as an AI provider for generating commit messages and PR titles, alongside existing providers like Anthropic and Copilot. It also refactors the provider output parsing logic for better maintainability and testability. The most important changes are grouped below:

Gemini Provider Integration:

• Adds a new GeminiProvider in internal/provider/gemini.go, enabling commit message and PR title generation using the Gemini CLI, including CLI invocation, prompt construction, and output parsing.
• Updates CLI commands and interactive config to support Gemini as a selectable provider, including default model and suggestion count, and ensures Gemini does not require an API key or endpoint. [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12]
• Ensures Gemini is properly referenced in the README and configuration defaults. [1] [2]

Provider Output Parsing Refactor:

• Extracts and centralizes output parsing logic into a new parseOutputLines function in internal/provider/common.go, which strips markdown, list prefixes, and skips empty/long lines.
• Adds comprehensive unit tests for parseOutputLines to ensure robust handling of various output formats.

Anthropic Provider Refactor:

• Refactors the Anthropic provider to use the new parseOutputLines function and move CLI execution into a reusable runCLI helper, reducing code duplication and improving maintainability. [1] [2]

Configuration and Defaults:

• Sets up default model and suggestion count for Gemini in the configuration system, and ensures endpoint handling is consistent with other CLI-based providers. [1] [2]

These changes collectively make it easy to use the Gemini CLI for AI-powered commit and PR title suggestions, while improving code clarity and maintainability.