Chore: Add unit and integ tests for already upgraded search functionality in jumpstart code

[mollyheamazon]

Issue #, if available:
Original code used eval() to evaluate user filter expressions in JumpStart search, allowing arbitrary code execution. This is already fixed and merged in: #5497

Description of changes:
Added 19 tests verifying the fix:

• Unit tests (18): Verify code injection patterns are blocked (__import__, exec, eval, file access, etc.) and AST parser is used instead of eval()
• Integration test (1): Confirms malicious queries don't execute in real API calls

All tests pass, confirming injection attempts return False without executing code.

Also added unit test fixes for sagemaker-train that is constantly failing for all PRs.

By submitting this pull request, I confirm that you can use, modify, copy, and redistribute this contribution, under the terms of your choice.