fix: blockdb file eviction race issue

[DracoLi]

Why this should be merged

Fix race for when a data file handler is evicted but another thread is still using that handler. The eviction tests caught this issue. See example, example 2

Resolves #4194

How this works

The eviction tests caught this since it simulated high evictions during concurrent calls. The simple solution here is to just retry opening the data file if we get an os.ErrClosed error during data file read/write, which should be rare.

How this was tested

Tests

[Copilot]

Pull Request Overview

This PR fixes a race condition in the blockdb file eviction system where a data file handler could be evicted while another thread is still using it, causing errors during concurrent operations.

• Implemented retry logic with os.ErrClosed detection for data file operations
• Updated getDataFileAndOffset to return the file index for eviction management
• Modified test cases to properly simulate file access errors

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 3 comments.

File	Description
x/blockdb/database.go	Added retry loops with os.ErrClosed handling for read/write operations and updated file access method signature
x/blockdb/database_test.go	Removed unnecessary type parameters from LRU cache instantiation
x/blockdb/writeblock_test.go	Updated test case to use file permission changes instead of file closure for error simulation

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[DracoLi]

the read body and header methods will be moved soon.