docs: align MiniLM truncation metadata

[Mirochill]

All Submissions:

• Have you followed the guidelines in our Contributing document?
• Have you checked to ensure there aren't other open Pull Requests for the same update/change?

Summary

• Align the supported-model description for sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2 with its 128-token default truncation length.
• Keep the model metadata in TextEmbedding.list_supported_models() consistent with the upstream model architecture noted in [Bug]: Docstring does not align with behaviour #531.

Fixes #531.

Validation

• git diff --check
• git show --check --stat --oneline HEAD
• Static search for the MiniLM model description in fastembed/text/pooled_embedding.py
• Not run locally: tests, docs build, pre-commit