EDGE-613 Fix Cognite client certificate auth

[Yaseen-A-Khan]

Background

This is in act of addressing one of the customer issues with cognite client certificate authentication failure where the error log showed, binascii.Error: Odd-length string, which happens when the thumbprint has an odd-length which was happening because the ConnectionConfig.get_cognite_client() was using bare str() on the incoming bytes from _load_certificate_data function.
Reference to the ticket: https://cognitedata.atlassian.net/browse/CDFBUG-1577

Summary

This PR adds the change to address the issue along with the unit and integration tests.

Test plan

• uv run pre-commit run --all passes (mypy, ruff, format)
• uv run pytest tests/tests_unit/test_certificate_auth.py -v — 7 unit tests pass

[gemini-code-assist]

Code Review

This pull request updates the return type annotation of _load_certificate_data to always return bytes, and updates the certificate loading logic to decode bytes to UTF-8 strings. The reviewer suggested using the more idiomatic .decode("utf-8") method instead of str(bytes, "utf-8") when converting the certificate thumbprint and key to strings.

[gemini-code-assist]

Using str(bytes, "utf-8") is unidiomatic in Python. It is highly recommended to use .decode("utf-8") instead, which is the standard and more readable way to convert bytes to a string.

Suggested change

	cert_thumbprint=str(thumprint, "utf-8"),
	certificate=str(key, "utf-8"),
	cert_thumbprint=thumprint.decode("utf-8"),
	certificate=key.decode("utf-8"),

[Yaseen-A-Khan]

as the functionality remains same for key.decode("utf-8") and str(key, "utf-8"), thinking str seemed to convey that it converts to a string better, i retained it.

[gemini-code-assist]

Using str(bytes, "utf-8") is unidiomatic in Python. It is highly recommended to use .decode("utf-8") instead, which is the standard and more readable way to convert bytes to a string.

Suggested change

	cert_thumbprint=str(thumbprint, "utf-8"),
	certificate=str(key, "utf-8"),
	cert_thumbprint=thumbprint.decode("utf-8"),
	certificate=key.decode("utf-8"),

[Yaseen-A-Khan]

as the functionality remains same for key.decode("utf-8") and str(key, "utf-8"), thinking str seemed to convey that it converts to a string better, i retained it.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 83.43%. Comparing base (3bb7bb5) to head (ea8fff6).
⚠️ Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #544      +/-   ##
==========================================
+ Coverage   83.08%   83.43%   +0.35%     
==========================================
  Files          45       45              
  Lines        4522     4522              
==========================================
+ Hits         3757     3773      +16     
+ Misses        765      749      -16     

Files with missing lines	Coverage Δ
cognite/extractorutils/__init__.py	100.00% <100.00%> (ø)
cognite/extractorutils/configtools/_util.py	77.04% <100.00%> (+11.47%)	⬆️
cognite/extractorutils/configtools/elements.py	81.77% <ø> (+1.40%)	⬆️
...te/extractorutils/unstable/configuration/models.py	85.71% <ø> (+0.85%)	⬆️

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[nithinb]

I know this is in draft now but a couple of things:

1. add integration test
2. include version bump in that so you can update the extractor right after this

[nithinb]

just two minor changes and a nit. Nit can be addressed later if you want

[nithinb]

risk review ok