Add MiniMax as a built-in LLM provider

[octo-patch]

Reason: Add MiniMax as a first-class built-in LLM provider.

This registers MiniMax as a native provider, wired through the same built-in
provider integration path already used by the other registered providers, so
MiniMax models can be selected and used out of the box.

Changes

• New provider wrapper src/pygpt_net/provider/llms/minimax.py, built on the
  shared LlamaIndex-compatible client and resolving api_key_minimax /
  api_endpoint_minimax (default https://api.minimax.io/v1).
• Register the provider in the launcher (src/pygpt_net/app.py).
• Register the provider id in the compatible-provider list
  (src/pygpt_net/core/types/openai.py) and in the chat client argument
  dispatch (src/pygpt_net/core/models/models.py).
• API-key presence check (src/pygpt_net/controller/chat/common.py) and
  model-importer environment wiring
  (src/pygpt_net/controller/model/importer.py).
• Default config keys api_endpoint_minimax and api_key_minimax
  (src/pygpt_net/data/config/config.json), a dedicated "MiniMax" settings tab
  with key and endpoint fields (src/pygpt_net/data/config/settings.json), and
  English labels (src/pygpt_net/data/locale/locale.en.ini).
• Two built-in MiniMax models in src/pygpt_net/data/config/models.json:
  • MiniMax-M3 — 1,000,000 token context; text, image and video input.
  • MiniMax-M2.7 — 204,800 token context; text input.

Missing config keys are added automatically on upgrade via the existing
post-upgrade config check.

Checks

• python -m py_compile on all changed Python modules — passed.
• JSON validity of config.json, settings.json, and models.json — passed.
• locale.en.ini parses cleanly.

The full test suite requires GUI and runtime dependencies that are unrelated to
this change and were not available in the build environment.