feat: add unk_token property to Unigram model

[ArthurZucker]

Summary

• Adds getter/setter for unk_token on the Unigram model
• Allows users to retrieve and modify the unknown token by its string value (in addition to the existing unk_id)

Changes

• Add get_unk_id(), get_unk_token(), set_unk_token() methods to Rust Unigram model
• Add Python bindings for unk_token and unk_id properties
• Add Python type hints for new properties
• Add comprehensive tests for Rust and Python

Example usage

from tokenizers.models import Unigram

vocab = [("<unk>", 0.0), ("hello", -1.0), ("world", -1.5)]
model = Unigram(vocab, unk_id=0)

# Get unk_token
print(model.unk_token)  # "<unk>"
print(model.unk_id)     # 0

# Set unk_token to a different token in vocab
model.unk_token = "hello"
print(model.unk_token)  # "hello"
print(model.unk_id)     # 1

Test plan

• Rust tests for get_unk_id, get_unk_token, set_unk_token methods
• Python tests for unk_token property getter
• Python tests for unk_token property setter
• Python tests for error handling (non-existent token, None value)

🤖 Generated with Claude Code

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.