Description

Add Voyage AI as a new embedding provider for grepai. Voyage AI offers code-optimized embedding models (e.g., voyage-code-3) that produce 1024-dimensional vectors, providing a cloud-based alternative tuned specifically for code search use cases.

This PR implements:

• Full VoyageAIEmbedder with batch embedding, adaptive rate limiting, token bucket throttling, and retry logic (including 429/5xx handling)
• Integration into the embedder factory (embedder.NewFromConfig / embedder.NewFromWorkspaceConfig)
• Interactive and non-interactive grepai init support (provider selection, endpoint/model defaults)
• Configuration defaults (DefaultVoyageAIEmbeddingModel, DefaultVoyageAIEndpoint, dimensions=1024)
• Correct dimension resolution in GetDimensions() for the voyageai provider
• Support across all code paths: CLI search, watch, workspace search, workspace watch, and MCP server

Related Issue

Fixes #

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation update
• Refactoring (no functional changes)
• Performance improvement
• Test update

How Has This Been Tested?

• Unit tests
• Integration tests
• Manual testing

Test Configuration:

• OS: Linux
• Go version: (project default)
• Embedding provider: voyageai (unit tests), ollama/openai (existing test suite)

Checklist

• My code follows the project's code style
• I have run golangci-lint run and fixed any issues
• I have added tests that prove my fix/feature works
• I have updated the documentation if needed
• I have added an entry to CHANGELOG.md (if applicable)
• My changes generate no new warnings
• All new and existing tests pass

Screenshots (if applicable)

N/A

Additional Notes

• The VOYAGE_API_KEY environment variable is used for authentication (falls back to config api_key field).
• Voyage AI dimensions default to 1024 (native for voyage-code-3), distinct from OpenAI's 1536 and the local embedders' 768.
• The embedder reuses the same batch/retry/rate-limiting infrastructure as the OpenAI embedder, with Voyage AI-specific request formatting (output_dimension field, input_type support).
• This branch was merged with the factory refactor branch; the conflict resolution preserved both Voyage AI support and the centralized embedder.NewFromConfig factory pattern, avoiding code duplication across CLI commands and MCP server.