[BUG] Dangling flint index when external scheduler operation failed

[noCharger]

What is the bug?
When scheduler operation failed. The index can still be created. This is due to the current implementation of optimistic transaction only rollback the log entry without clean up the resources

How can one reproduce the bug?
Steps to reproduce the behavior:

1. Go to '...'
2. Click on '....'
3. Scroll down to '....'
4. See error

What is the expected behavior?
A clear and concise description of what you expected to happen.

What is your host/environment?

• OS: [e.g. iOS]
• Version [e.g. 22]
• Plugins

Do you have any screenshots?
If applicable, add screenshots to help explain your problem.

Do you have any additional context?
Add any other context about the problem.

[andrross]

[Catch All Triage - 1, 2, 3, 4, 5]