The token normalization logic treated empty string '' and empty list [] inconsistently. Empty string loaded from param_manager (env var/config file), while empty list filtered to None directly without checking param_manager.

Changes

• Refactored token normalization to use extract_list_of_str consistently, treating None, '', and [] identically - all load from param_manager
• Added test coverage for empty list and empty string loading from environment variables
• Fixed test naming for pytest discovery

Behavior

Before:

configure(token='')   # Loads from LOGFIRE_TOKEN env var ✓
configure(token=[])   # Returns None, ignores env var ✗

After:

configure(token='')   # Loads from LOGFIRE_TOKEN env var ✓
configure(token=[])   # Loads from LOGFIRE_TOKEN env var ✓

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.