Fix, use filtered valid_questions for language detection (closes #76)

[OscarAR46]

Description

Fixed a bug where the 'detect()' func was receiving unfiltered data instead of the filtered 'valid_questions' list, causing 'LangDetectException: No features in text' errors when processing Excel files with empty or invalid question text.

Fixes #76

Type of change

• Bug fix (non-breaking change which fixes an issue
• Only one line changed in this one file

Testing

• Verified that the fix passes the filtered 'valid_questions' to 'detect()' instead of the raw 'df_questions["question"]'
• No new dependencies added

Test Configuration

• Library version: Latest main branch
• OS: Windows
• Toolchain: Python 3.x

Checklist

• My PR is for one issue, rather than for multiple unrelated fixes.
• My code follows the style guidelines of this project.
• I have performed a self-review of my own code
• My changes generate no new warnings
• The Harmony API is not broken by my change to the Harmony Python library

I hope this PR checklist suffices! Simple fix so did not include all checklist criteria :)

[woodthom2]

Thanks Oscar!