fix: ensure valid UTF8 byte sequence for inference history

[mattjohnsonpint]

Fixes #903

[Copilot]

Pull Request Overview

This PR ensures that any byte sequence produced for inference history is valid UTF-8 by sanitizing it before storage.

• Refactored getInferenceDataJson to collect serialized bytes in a single variable.
• Added a call to utils.SanitizeUTF8 before returning the result.
• Updated CHANGELOG with the new fix entry.

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
runtime/db/inferencehistory.go	Unified byte handling in getInferenceDataJson and applied UTF-8 sanitization
CHANGELOG.md	Added entry for the UTF-8 sanitization fix for inference history

Comments suppressed due to low confidence (3)

runtime/db/inferencehistory.go:164

• [nitpick] Consider renaming the local variable bytes to avoid shadowing the bytes package and improve readability (e.g., data or buf).

	var bytes []byte

runtime/db/inferencehistory.go:178

• Add unit tests to verify that invalid UTF-8 byte sequences are correctly sanitized by SanitizeUTF8, ensuring coverage for this new behavior.

	return utils.SanitizeUTF8(bytes), nil

runtime/db/inferencehistory.go:160

• Update the function comment to mention that output is now sanitized for valid UTF-8 sequences using SanitizeUTF8.

	// If the value is a byte slice or string, it must already have been serialized as JSON.