feat: Add Google Gemini support to 01-foundations/02-prompt-engineering

[AmeyaMprojects]

This pull request adds native support for Google Gemini models across the entire 02-prompt-engineering foundational module. It expands the repository's educational coverage by demonstrating how first-principles prompting layered concepts (system configuration, few-shot structures, and native JSON constraints) map to the official, modern google-genai SDK.
Changes Included

03_system_prompts_gemini.py: Parallels the existing Anthropic and OpenAI implementations to demonstrate role engineering, progressive constraint adjustment, and system instruction patterns using types.GenerateContentConfig(system_instruction=...).

06_few_shot_cot_gemini.py: Implements zero-shot, custom label few-shot classification, and sequential reasoning via Chain-of-Thought (CoT) setups.

09_structured_output_gemini.py: Compares manual, prompted extraction schemas against Gemini’s strict, native schema enforcement using Pydantic models fed directly into response_schema.

common.py: Integrated a robust token tracker class (GeminiTokenTracker) to capture and expose upstream billing details (prompt_token_count / candidates_token_count) seamlessly alongside existing provider trackers.

README.md: Updated to complete the sequential 1-9 numbering of files, adding architectural script snippets and updating comparative provider tables.

Testing & Verification

All scripts have been tested locally via uv run environments using a gemini-2.5-flash model target under zero-temperature configurations.

System routing and conditional menus match interactive performance logs identically.

Structural JSON calls passed string extraction clean tests and verified completely against strict native schema blocks.

[AmeyaMprojects]

Let me know if this is fine so i can work on other folders