feat: implement real API for kpi_api_analytics.py

[Nidhi1202]

Closes #160

Summary

Implements the KPI Analytics Lambda with real PostgreSQL queries against the Virginia RDS database.

What's implemented

• Request Status Distribution — counts requests joined to the request_status lookup, filtered to the dashboard statuses (CREATED, IN_PROGRESS, RESOLVED)
• Total Requests — count across all statuses
• Average Resolution Time by Category — AVG(serviced_date - submission_date) in hours, joined to help_categories for the category name, for RESOLVED requests only
• SLA metadata — always returned, per the dashboard spec

Notes

• The request_status lookup table contains RESOLVED but not COMPLETED, so the resolution-time query filters on RESOLVED only.
• The issue text states the formula as submission_date - serviced_date, which would produce negative values. Implemented as serviced_date - submission_date.
• Added an optional environment-variable path for the DB connection so the Lambda can be run locally without AWS/SSM access. Production still reads credentials from SSM.

Testing

Returns the safe default response when credentials are not configured. Verified locally against a PostgreSQL instance built from the schema in Saayam_Table.column.names_data.xlsx — statusCode 200, all four keys present, empty DB returns [] and 0 without crashing, and the connection closes in the finally block. Full integration testing requires Virginia RDS credentials.