fix: Update benchmark script to use process_text API (#51)

[ada-cinar]

Problem

The benchmark script was using the old string-based Pipeline API:

pipeline = durak.Pipeline(["clean", "tokenize", "remove_stopwords", "normalize"])

This caused a TypeError: 'str' object is not callable because the Pipeline class expects callable objects, not strings.

Solution

Updated benchmark #4 to use the process_text convenience function instead:

def pipeline_func(text):
    return durak.process_text(text, remove_stopwords=True, rejoin_suffixes=True)

This uses the current API and should run without errors.

Testing

• ✅ Syntax validation passed (python3 -m py_compile)
• 🔄 Full benchmark run requires maturin develop (CI will verify)

Closes #51