feat: upgrade MiniMax default model to M3

[octo-patch]

Summary

Upgrade MiniMax model configuration to use M3 as the default model.

Changes

• Set MiniMax-M3 as the default model in mem0/llms/minimax.py
• Keep MiniMax-M2.7 as a supported alternative (no breaking change for users explicitly selecting M2.7)
• Update the default-model unit test to assert MiniMax-M3
• Add a new test verifying the factory default model is MiniMax-M3
• Update the usage doc to reflect M3 as the default

Why

MiniMax-M3 is the latest flagship model, with a 512K context window, up to 128K output, and image input support. Setting M3 as the default gives new users the most capable model out of the box while users on M2.7 continue to work by explicitly setting model: "MiniMax-M2.7".

Testing

• All 10 unit tests in tests/llms/test_minimax.py pass locally
• Default-model assertion updated to MiniMax-M3
• Existing M2.7 tests preserved to verify M2.7 still works when explicitly selected

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.

---

octo-patch seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}