Add Inkling deployment helper for Akash

[Tibo2403]

What changed

• adds scripts/bash/deploy_inkling_akash.sh
• generates an Akash SDL targeting 1x RTX 3090 by default
• deploys a LiteLLM gateway to a remote OpenAI-compatible Inkling endpoint
• explicitly blocks misleading self-host deployment on a single RTX 3090
• adds docs/inkling-akash.md with setup, security, cost, and test instructions

Why

Inkling cannot be self-hosted on a 24 GiB RTX 3090. The helper preserves the requested Akash GPU option while making the limitation explicit and providing a usable API-gateway deployment path.

Validation

• script uses strict Bash mode and validates numeric/options input
• --dry-run prints the generated SDL without writing or deploying
• deployment requires explicit --deploy and an AKASH_KEY_NAME
• generated files containing credentials are created with restrictive permissions

Important limitation

The RTX 3090 is not used for inference in API mode; a CPU-only Akash gateway would be cheaper. The documentation calls this out explicitly.