fix(library): use bot refuse to respond in gliner PII detection flows

[Pouyanpi]

Description

GLiNER PII detection was using "bot inform answer unknown" which returned a misleading "I don't know" message. Replaced with "bot refuse to respond" to correctly indicate an intentional block, consistent with all other rails.

[greptile-apps]

Greptile Summary

This PR correctly replaces bot inform answer unknown ("I don't know") with bot refuse to respond ("I'm sorry, I can't respond to that.") in GLiNER PII detection flows. This semantic improvement clarifies that PII detection is an intentional block rather than expressing uncertainty.

Key changes:

• Updated three detection flows in both flows.co and flows.v1.co (input, output, retrieval)
• Removed custom test definitions and updated expected messages to match the standard refusal message
• Aligns with regex detection library pattern

Note: Other PII detection libraries (privateai, pangea, patronusai) still use bot inform answer unknown. Consider standardizing these in a follow-up PR for consistency across all PII detection implementations.

Confidence Score: 5/5

• This PR is safe to merge with no issues detected
• The changes are straightforward, semantically correct, and properly tested. The refactoring improves message clarity by replacing "I don't know" with "I can't respond to that" for PII blocking scenarios. All test assertions are correctly updated to match the new standard message.
• No files require special attention

Important Files Changed

Filename	Overview
nemoguardrails/library/gliner/flows.co	Replaced bot inform answer unknown with bot refuse to respond in three PII detection flows (input, output, retrieval) - semantically correct change
nemoguardrails/library/gliner/flows.v1.co	Same changes as flows.co but for v1 syntax - replaced bot inform answer unknown with bot refuse to respond in three subflows
tests/test_gliner.py	Removed custom bot inform answer unknown definitions and updated test assertions to expect "I'm sorry, I can't respond to that." message

_{Last reviewed commit: 024aa8d}

[greptile-apps]

_{3 files reviewed, no comments}

_{Edit Code Review Agent Settings | Greptile}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[tgasser-nv]

Looks good, do you need to add define bot refuse to respond at the top of the Colang flows.co file like the content-safety, topic-safety, etc?

library $ find . -name 'flows.v1.co' -exec grep -Hn 'define bot refuse to respond' {} \;
./jailbreak_detection/flows.v1.co:1:define bot refuse to respond
./llama_guard/flows.v1.co:1:define bot refuse to respond
./content_safety/flows.v1.co:1:define bot refuse to respond
./regex/flows.v1.co:1:define bot refuse to respond
./topic_safety/flows.v1.co:3:define bot refuse to respond
./fiddler/flows.v1.co:1:define bot refuse to respond
./autoalign/flows.v1.co:37:define bot refuse to respond
./self_check/input_check/flows.v1.co:1:define bot refuse to respond
./self_check/output_check/flows.v1.co:1:define bot refuse to respond
./self_check/facts/flows.v1.co:1:define bot refuse to respond

Also could you do a local integration test (instructions are here) to make sure there are no errors and the response is correct?