Fix MovieLens dataset incompatibility with SentenceTransformers =5.x

[drivanov]

Summary

This PR fixes a runtime error in the MovieLens dataset when used with newer versions of sentence-transformers (≥5.x).
The issue arises from passing a NumPy array (df['title'].values) to SentenceTransformer.encode(), which leads to incorrect modality inference and raises:

ValueError: Modality 'audio' is not supported by this SentenceTransformer model.

Root Cause

Recent versions of SentenceTransformers` introduced modality-aware input handling.
When a NumPy array is passed:

df['title'].values  # np.ndarray(dtype=object)

the library may misinterpret the input as a non-text modality (e.g., audio), instead of text.
However, encode() expects a List[str] for text inputs.

Fix

Convert the input to an explicit list of strings before encoding:

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 84.18%. Comparing base (c211214) to head (b988ba9).
⚠️ Report is 194 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master   #10668      +/-   ##
==========================================
- Coverage   86.11%   84.18%   -1.93%     
==========================================
  Files         496      510      +14     
  Lines       33655    36022    +2367     
==========================================
+ Hits        28981    30325    +1344     
- Misses       4674     5697    +1023     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.