feat: add MCP prompts for common GA4 workflows (closes #52)

[ZLeventer]

Summary

Adds MCP Prompts to the server, closing #52. Prompts are exposed as slash commands in Gemini CLI and as pre-built prompts in Claude Desktop and other MCP clients.

Prompts added

Prompt	Use case
traffic-summary	Aggregate sessions, users, bounce rate, and engagement time for a date range
top-pages	Top N pages by sessions with engagement metrics
acquisition-overview	Traffic broken down by source / medium, grouped by channel
compare-periods	Side-by-side metric comparison between two date ranges with % change
campaign-performance	UTM-tagged campaigns ranked by conversions
realtime-overview	Active users now by page, country, and device (three run_realtime_report calls)

All prompts accept property_id (required) and most accept an optional date_range argument.

Implementation

• analytics_mcp/prompts.py (new) — prompt definitions (mcp_types.Prompt list) and text builders. Each builder returns a GetPromptResult with a user-role message that instructs the model which tools to call and how to present results.
• analytics_mcp/coordinator.py — adds @app.list_prompts() and @app.get_prompt() handlers that delegate to prompts_module.
• tests/prompts_test.py (new) — 16 unit tests covering list_prompts and get_prompt for every prompt, including default argument handling and unknown prompt error.
• README.md — new Prompts section with a table of all prompts and their arguments.

Tests

16 passed in 0.34s