Add mcs-finance-statement-agent sample

[sgshaji]

Summary

Adds mcs-finance-statement-agent — a Microsoft Copilot Studio sample for credit financial statement extraction.

An analyst uploads a PDF financial report in Microsoft Teams; a Copilot Studio agent orchestrates an Azure Functions pipeline that extracts the Income Statement, Balance Sheet, Cash Flow, and computed Ratios into a downloadable Excel workbook ready for credit spreading.

Highlights

• Async polling pattern to handle the 30-second custom-connector timeout for long-running extraction jobs (HTTP 202 + jobId, polled every 30 s by a ConditionGroup + GotoAction loop)
• Multi-language label reconciliation (EN / CN / JP / FR) via Azure OpenAI GPT-4.1
• Managed Identity end-to-end for Azure service-to-service auth — no API keys
• Pluggable extraction backend — Content Understanding (default), Document Intelligence, Textract, or local pdfplumber
• Optional HITL review Code App — React + TypeScript, backed by Dataverse
• Job state in Blob Storage with 30-min TTL; SAS URLs returned in chat for the generated Excel

Folder structure

samples/mcs-finance-statement-agent/
├── README.md
├── .gitignore
├── assets/architecture.png
└── src/
    ├── azure-functions/         # Python backend (HTTP 202 async pipeline)
    ├── copilot-studio-agent/    # Copilot Studio YAML
    ├── code-app/                # Power Apps Code App (HITL grid)
    └── docs/

Checklist

• Folder name uses mcs- prefix per CONTRIBUTING
• README.md based on _SAMPLE_templates/any-sample
• assets/ contains the screenshot referenced in README
• .gitignore included
• Tracking pixel updated to samples/mcs-finance-statement-agent
• No API keys, secrets, or tenant-specific values committed
• Author block includes name, GitHub handle, company

Test plan

• README renders correctly on GitHub (architecture diagram visible)
• All Minimal-Path-to-Awesome steps work on a fresh tenant + Azure subscription
• End-to-end: upload sample PDF → Excel link returned in Teams chat