feat: make blob v2 dedicated threshold configurable

[yanghua]

No description provided.

[github-actions]

Code Review

P1 - Thread Safety Issue in Test

The test test_dedicated_threshold_env_parsing_only manipulates environment variables using std::env::set_var and std::env::remove_var. Since Rust tests run in parallel by default, this can cause race conditions with other tests that may also read environment variables or use get_dedicated_threshold().

Suggested fix: Either:

1. Add #[serial] attribute using the serial_test crate to ensure this test runs in isolation
2. Or use a test-specific approach that doesn't rely on global env state (e.g., test the parsing logic in isolation with a helper that takes the value as a parameter)

---

The overall implementation approach is reasonable. The threshold is read once at BlobPreprocessor construction time, which avoids repeated env lookups during processing.

[Xuanwo]

Thank you for working on this, I hope this can be a column metadata like blob_dedicated_size_threshold instead.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.

📢 Thoughts on this report? Let us know!

[yanghua]

Thank you for working on this, I hope this can be a column metadata like blob_dedicated_size_threshold instead.

OK, got it.

[Xuanwo]

Others LGTM

[yanghua]

@Xuanwo I have addressed your comments.

[Xuanwo]

Thank you for working on this!