[FEATURE] Implement Local RAG Pipeline

[FrilLab]

Problem

Even with file context and symbol search, large projects contain more information than can fit into a single prompt.

Important context may be missed.

Proposed Solution

Implement a local Retrieval-Augmented Generation (RAG) pipeline.

Possible workflow:

Workspace
↓
Chunking
↓
Embedding
↓
Vector Search
↓
Relevant Context
↓
LLM Translation

Retrieved context should be included automatically when generating translations and explanations.

Alternatives

• Large prompt windows
• Full project context injection

Additional Context

This represents the long-term intelligence layer of Frilingo.

Goal:

Translate code and documentation using knowledge retrieved from the user's actual project.