Why dubbo does choose to create a lock file and use nio file lock

- [x] I have searched the [issues](https://github.com/apache/dubbo/issues) of this repository and believe that this is not a duplicate.

## Ask your question here
Why dubbo does choose to create a lock file and use nio file lock to control the concurrent problem during saving cache properties to disk instead of other locks? Thanks.

Start by locating Dubbo's cache-properties persistence code and the use of Java NIO file locks. Read the surrounding concurrency and disk-saving logic, then compare the lock's role with the alternatives discussed in the issue; done means documenting a clear rationale for this design.

Written by the indexing model from the issue text.