[Context Provider] Add Dakera Memory context provider — persistent semantic recall across coding sessions

Summary

This issue proposes adding a Dakera Memory context provider that allows Continue users to semantically recall relevant coding memories and past context via @dakera in the chat interface.

Problem

As developers work across sessions, important context gets lost — architectural decisions, API contracts explored, implementation notes, and past debugging sessions. Today there is no built-in way to inject that long-term context into Continue conversations.

Solution

Dakera is a self-hosted vector memory server that stores and retrieves developer context using semantic search with decay-weighted importance scoring. A DakeraContextProvider would let Continue users query their personal or team memory store directly from the chat input.

Example usage:

@dakera what did I implement last week in the auth module?
@dakera how did we handle rate limiting in the API?

Continue surfaces the most relevant stored memories as context items, which are then injected into the next LLM call alongside the user's question.

Proposed API

The provider calls POST /v1/memories/search on the user's Dakera instance:

{
  "query": "auth module implementation",
  "top_k": 10
}

Results are mapped to ContextItem[] with the memory text as content.

Config example (config.json)

{
  "contextProviders": [
    {
      "name": "dakera",
      "params": {
        "baseUrl": "http://localhost:3000",
        "apiKey": "optional-key",
        "topK": 10,
        "sessionId": "optional-session-filter"
      }
    }
  ]
}

Why it fits the Continue ecosystem

• Follows the same pattern as HttpContextProvider and other third-party integrations
• Fully self-hosted — no external cloud dependency required; works air-gapped
• Complements codebase search (@codebase) with persistent, cross-session developer memory
• Optional apiKey and sessionId support for team and per-project isolation

I have an implementation ready in a PR.

Start by comparing the proposed DakeraContextProvider with HttpContextProvider and verify the POST /v1/memories/search mapping into ContextItem[]. Check the config.json example for baseUrl, apiKey, topK, and sessionId; done means @dakera queries return memory content in chat with the stated filters.