Fix DbtVirtualenvBaseOperator to use correct virtualenv Python path

[kesompochy]

Description

This PR addresses an issue where the DbtVirtualenvBaseOperator was executing dbt commands using the system-wide Python path instead of the virtualenv path. The root cause was that the self reference in the run_subprocess method was bound to a different instance than the one created during initialization, likely due to Airflow's DAG pickling mechanism.
To resolve this, we've refactored the invoke_dbt and handle_exception methods to be properties. This ensures that they dynamically reference the correct method of the current instance at runtime, rather than being bound to a potentially stale instance from initialization.

Related Issue(s)

fix #1246

This may be related to issue #958 in version 1.5.0

Breaking Change?

No

Checklist

• I have made corresponding changes to the documentation (if required)
• I have added tests that prove my fix is effective or that my feature works

Additional Notes

I acknowledge that ideally, a test should be added to reproduce the original issue and verify the fix. However, I found it challenging to create an appropriate test, especially considering that this might require an integration test with Airflow to properly simulate Airflow behavior. If there are suggestions for how to effectively test this scenario, I would greatly appreciate the guidance.

I sincerely apologize for introducing this bug in the first place with PR #1200. I understand this has caused inconvenience, and I'm grateful for the opportunity to fix it. I kindly request a thorough review of these changes to ensure we've fully addressed the issue without introducing new problems.

[netlify]

✅ Deploy Preview for sunny-pastelito-5ecb04 canceled.

Name	Link
🔨 Latest commit	25f75ee
🔍 Latest deploy log	https://app.netlify.com/sites/sunny-pastelito-5ecb04/deploys/671238e96dbc1e0008bb23ae

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 95.73%. Comparing base (ae4ec78) to head (25f75ee).
Report is 9 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1252   +/-   ##
=======================================
  Coverage   95.73%   95.73%           
=======================================
  Files          67       67           
  Lines        3965     3967    +2     
=======================================
+ Hits         3796     3798    +2     
  Misses        169      169           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[tatiana]

HI @kesompochy thank you very much for working on this.

When we write code, there is always the risk we'll be introducing bugs, and in this particular case, the reviewer is as accountable as the author. So, please, don't blame yourself.

As you mentioned, I really believe we need to write an integration test to try to reproduce the original issue - or a scenario closer to it, that fails before, and passes with this fix.

I'll take sometime to reproduce the issue and try to give ideas on how we could accomplish this. We'll be releasing this in a patch release of Cosmos, if not 1.7.1, then 1.7.2.

[tatiana]

@kesompochy I've added a test more representative of the change in #1286. I'll merge your PR and the test I created to validate this change.

[kesompochy]

@tatiana, Thank you for reviewing and for adding such a comprehensive test case. Using dag.test() in the integration test (#1286) is an excellent way to reproduce the pickling issue in a controlled environment.
Thanks to your help and the community's support, I'm glad the fix for the bug I introduced was properly verified. Really appreciate your thorough review!