Add SecretsDetector with entropy and prefix heuristics and integrate into CLI scan

[Jeffrin-dev]

Motivation

• Provide secret-detection support alongside PII scanning to surface API keys and other high-entropy tokens during scan CLI runs.
• Use both an optional detect-secrets library and in-house heuristics (Shannon entropy, prefix patterns, and stopword filtering) to reduce false positives.

Description

• Add shadowaudit.core.secrets.SecretsDetector implementing entropy scoring, candidate extraction, prefix pattern matching, optional detect-secrets integration, and filtering heuristics.
• Integrate SecretsDetector into the scan CLI command to populate secrets_found on the scan result and set action_taken to detected when either entities or secrets are present.
• Include explicit prefix patterns for common token formats (e.g., sk-, ghp_, AKIA, xoxb-) and a stopword list plus minimum candidate length to avoid flagging ordinary words.
• Add and update unit tests to cover entropy scoring, prefix pattern detection, ignoring normal English sentences, and CLI-level secret detection.

Testing

• Ran the test suite with pytest including the new tests in tests/test_cli_scan.py and tests/test_policy_and_secrets.py.
• All automated tests, including the new secrets detection and CLI tests, passed successfully.

---

Codex Task