[lucasjia-aws]

Summary

DeserializerWrapper currently forces its configured Accept type for both request and response deserialization. As a result, valid JSON or CSV requests can be incorrectly parsed as NPY data.

This change makes input deserializers honor the caller-provided Content-Type while preserving the existing output behavior that uses the configured Accept type when model servers return misleading response headers such as text/html.

Testing

• Added NumPy input regressions for JSON and CSV payloads.
• Added output coverage proving that a misleading text/html response still uses the configured NPY deserializer.
• Ran the affected schema-builder suite: 8 passed.
• Black, Flake8, security exposure checks, and git diff --check passed.
• Built the sagemaker-serve wheel and source distribution successfully.

Related change

The companion V2 fix is published on #6463.

Fixes #5006

[github-actions]

🤖 AI code review

This PR fixes DeserializerWrapper so that input deserializers honor the caller-provided Content-Type (via a new honor_content_type flag) while output deserializers keep using the configured Accept type, to work around model servers that return misleading response headers like text/html. The change is small, backward compatible (falls back to self._accept[0] when content_type is None or when the flag is false), and the added unit tests cover the JSON/CSV input cases and the misleading-output case. No issues found.

_{Reviewed commit 9abb543. Automated review; verify before acting.}