[Nithin (nithin42)]

Description

Closes #3468

Adds three common cloud and AI platform credential patterns to CredentialLeakScorer in both _DEFAULT_PATTERNS and GARAK_PATTERNS (following the same pattern structure and boundary checks as #3286):

• Azure SAS Token: detects SAS query strings (sv=20\d{2}-\d{2}-\d{2}&...&sig=...) with URL-encoded or raw base64 signatures.
• Azure Service Bus / Event Hub Key: detects SharedAccessKey=... values in semicolon-delimited Azure connection strings while ignoring SharedAccessKeyName.
• Databricks API Token: detects dapi personal access tokens (dapi[0-9a-f]{32} with optional shard suffix -[0-9]).

Each entry in GARAK_PATTERNS wraps the secret in (?P<credential>...) so CredentialLeakScorer.from_excluded_values(...) can exclude known test fixtures cleanly.

cc Roman Lutz (@romanlutz) (since this follows up on #3286 for Azure and Databricks credentials — would appreciate a review when you have a moment!)

Tests and Documentation

• Added unit tests in tests/unit/score/regex/test_credential_leak_scorer.py covering positive matches, short/malformed false positives, overlong token rejection, non-ASCII surrounding text, and from_excluded_values with GARAK_PATTERNS.
• Updated pattern count checks in tests/unit/score/regex/test_credential_leak_scorer.py and tests/unit/scenario/garak/test_api_key.py.
• Ran pytest tests/unit/score/regex/ tests/unit/scenario/garak/test_api_key.py locally (all passing).

[Nithin (nithin42)]

Hi Roman Lutz (@romanlutz), since this follows up on #3286 to cover Azure SAS tokens, SharedAccessKey connection strings, and Databricks PATs in CredentialLeakScorer, I'd appreciate your review whenever you have a moment. Thanks!