Add Sarvam AI integration for AI-171 mobile STT & TTS evaluation

[gyanLM10]

Add Sarvam AI (native Indic STT & TTS) as the foundation for evaluating multilingual STT/TTS support on mobile devices.

---

Jira Ticket: https://mifosforge.jira.com/browse/AI-171 Evaluate multilingual support for Sarvam AI

---

Implementation of Sarvam Saaras & Bulbul APIs:

• Mobile focus: Fast API endpoints, fully unblocks mobile clients (iOS/Android) from local storage/VRAM bottlenecks.
• Performance: Resolves Out-Of-Memory (OOM) crashes on 8GB machines and provides sub-second inference latency critical for edge devices.
• Tooling: Dedicated run_multilingual_eval.py script for measuring end-to-end pipeline inference speed and accuracy.
• Edge optimization: Starter foundation for offloading heavy native TTS/STT generation to optimized cloud inference for Indic languages.

---

What's included:

• Sarvam_Integration/: Directory isolating the integration experiments and Langchain tools.
• Sarvam_Integration/VoiceBanking_Results/: Submodule containing the Sarvam STT & TTS implementation scripts.
• Sarvam_Integration/VoiceBanking_Results/run_multilingual_eval.py: Minimal wrapper evaluation script.
• Sarvam_Integration/VoiceBanking_Results/RESULTS.md: Documentation hub for findings & reproduction.

---

⚙️ Development & Code Accountability

• AI Model Used: Gemini 2.5 Pro (via Google Deepmind)

---

---

Full Accountability:
All code in this PR remains under complete human (my) responsibility.

---

How to Reproduce:

git clone https://github.com/gyanLM10/community-ai.git
cd community-ai
git checkout feature/sarvam-integration-only
cd Sarvam_Integration/
python -m venv .venv && source .venv/bin/activate
pip install -r requirements.txt
cd VoiceBanking_Results
echo "SARVAM_API_KEY=your_key" > .env
python run_multilingual_eval.py