[good first issue-platform adapt-gp.nvim] Add Memory adapter for gp.nvim

[LeonSGP43]

Summary

Adds a docs-only adapter for gp.nvim (Robitx/gp.nvim, the GPT plugin for Neovim with chat/command/inplace workflows), connecting it to the TencentDB Agent Memory proxy through its native OpenAI-compatible providers table.

• adapters/gp-nvim/README.md + README_CN.md (EN/CN, same structure as the avante.nvim / codecompanion.nvim docs-only adapters)
• No code changes: the provider entry (endpoint, secret) and the agent link (provider = ...) live in the user's own Neovim config, so no files ship with the adapter.

How it connects (verified against official sources)

• gp.nvim supports "Any other 'OpenAI chat/completions' compatible endpoint" via the providers table (official README, Multiple providers section); the built-in ollama entry (endpoint = "http://localhost:11434/v1/chat/completions", no secret) is the canonical local-endpoint example.
• Unlike base-URL clients, gp.nvim's endpoint is the full chat completions URL — the adapter points it at the proxy's canonical route http://127.0.0.1:8096/codebuddy/<spaceId>/v1/chat/completions.
• secret supports os.getenv(...) for API keys (same capability as the legacy openai_api_key field per the README), so the sk-mem-... business user key is sent as Authorization: Bearer without landing in dotfiles.

Contents

• Prerequisites (one-command stack, business sk-mem-... user key)
• Setup: export key → declare provider + agent → align model with PROXY_UPSTREAM_MODEL → verify via :GpChatNew and the session picker
• Configuration reference table + troubleshooting table (404/key/model-id/session-picker/GUI-env rows)
• Notes: docs-only rationale, chat/command coverage, data-flow statement

Part of the platform-adapt batch tracked in #926.

[Maxwell-Code07]

Thank you so much for your attention and contribution! We will arrange an internal review for this PR shortly, and all feedback will be shared right here in the discussion.