Fix CWE-918 SSRF in chat-with-index/setup_env.py

[ayushhgarg-work]

Description

Fixes CWE-918 (Server-Side Request Forgery) in sdk/python/generative-ai/rag/code_first/flows/chat-with-index/setup_env.py, flagged by CodeQL under the Glasswing Mythos - July campaign (IcM 841731124).

The Promptflow setup_env tool previously copied connection.api_base directly into os.environ["OPENAI_API_BASE"], allowing an untrusted URL to flow to an HTTP sink. This PR adds _validate_api_base() which validates the URL before it is written to the environment:

• Must be a non-empty str
• No whitespace or control characters (closes urlparse leading-whitespace bypass, CRLF injection, null-byte tricks)
• https scheme only
• Hostname must match a static allow-list: *.openai.azure.com, *.cognitiveservices.azure.com, *.api.cognitive.microsoft.com, api.openai.com

Validated against a 52-case adversarial suite locally: all 8 legitimate URL patterns accepted; all 44 SSRF vectors (IMDS 169.254.169.254, loopback, private ranges, suffix collision e.g. x.openai.azure.com.evil.com, userinfo tricks api.openai.com@evil.com, fragment/query tricks, scheme downgrades) blocked with ValueError.

No behavior change for real Azure OpenAI / OpenAI endpoints.

Checklist

• I have read the contribution guidelines.
  • General
  • SDK
  • CLI
• I have coordinated with the docs team (mldocs@microsoft.com) if this PR deletes files or changes any file names or file extensions. — N/A, no file/path changes.
• Pull request includes test coverage for the included changes. — Validator verified against a 52-case adversarial suite locally (see PR description); existing sdk-generative-ai-rag-code_first-flows-chat-with-index-src-test.yml CI workflow exercises the sample end-to-end.
• This notebook or file is added to the CODEOWNERS file, pointing to the author or the author's team. — File already covered by existing CODEOWNERS entry for sdk/python/generative-ai/rag/.