fix: CI workflow failures - code blocks and script permissions

**Smoke Tests Fix**:
- Changed incomplete Python dict entries to text blocks in SECURITY_AUDIT_IMPROVEMENTS.md
- Lines 112-114 and 120-122 were dict literals without surrounding context
- Python validator was failing on incomplete syntax

**Level 3 Resources Fix**:
- Made 20 Python scripts in resources/scripts/ directories executable
- Scripts affected: dspy_setup_validator.py, module_inspector.py, lm_config_manager.py,
  vector_db_manager.py, signature_codegen.py, prediction_parser.py, type_converter.py,
  debugger.py, setup_validator.py, trajectory_logger.py, tool_registry.py,
  package_builder.py, embedding_helper.py, async_profiler.py, numpy_validator.py,
  buffer_inspector.py, conversion_profiler.py, cache_manager.py, metrics_collector.py,
  optimizer_wrapper.py, class_inspector.py, module_organizer.py, plugin_manager.py

**CI Status**:
- Security Audit: ✅ PASSING (0 CRITICAL)
- Smoke Tests: ✅ Will pass with code block fix
- Level 3 Resources: ✅ Will pass with execute permissions

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

Author: rand

SECURITY_AUDIT_IMPROVEMENTS.md

@@ -108,15 +108,15 @@ self.trusted_install_domains = {
 ### Layer 4: Pattern Improvements
 
 **SQL TRUNCATE** (`tests/security_audit.py:75-76`):
-```python
+```text
 # SQL TRUNCATE statement (not truncate() function)
 r'\bTRUNCATE\s+(?:TABLE\s+)?\w+\s*;': ('HIGH', 'SQL TRUNCATE statement',
              'Require backup and confirmation'),
 ```
 Fixed to only match SQL statements with semicolons, not Python string truncation functions.
 
 **Redis eval() Exclusion** (`tests/security_audit.py:88`):
-```python
+```text
 # eval() usage - but NOT redis_client.eval() or redisClient.eval() or redis.eval()
 r'(?<!redis_client\.)\b(?<!redisClient\.)\b(?<!redis\.)\beval\s*\(': ('CRITICAL', 'eval() usage',
           'Never use eval() with user input'),