fix(intelligence): keep fresh zero-importance memories searchable

[wayyoungboy]

Summary

• Prevent freshly persisted zero-importance memories from receiving zero initial/current retention
• Add a regression unit test covering relevant search hits with importance_score=0
• Format the touched intelligence files with black

Root Cause

#1084 made search ranking multiply storage relevance by effective_retention. Fresh memories whose importance evaluator returned 0.0 were initialized with current_retention=0.0, which zeroed out otherwise relevant hybrid-search results. This caused main Regression Tests to fail in test_native_hybrid_search.py::TestNativeHybridSearch::test_tc005_hybrid_search_fusion_effect, where a Zhang San hit was demoted below an unrelated Wang Wu result.

Verification

• pytest tests/unit/intelligence/test_retention_runtime.py tests/unit/intelligence/test_ebbinghaus_decay_rate.py tests/unit/intelligence/test_intelligent_memory_manager.py
• uv run --no-project --python 3.11 --with-editable ".[dev]" black --check src/powermem/intelligence/ebbinghaus_algorithm.py tests/unit/intelligence/test_retention_runtime.py
• git diff --check

Release Note

This should be merged before cutting the next release because latest main currently has a failing Regression Tests workflow.