The Metastore lock timeout led to the loss of metadata overwrite data.

### Apache Iceberg version

1.10.1 (latest release)

### Query engine

Spark

### Please describe the bug 🐞

1. We have a third-party service to clean up snapshot versions. During the cleaning process, if there is a long waiting time after the first confirmation of the lock's survival as shown;
2. During the waiting period, there were other normal submissions of batch 2026-02-08, and metadata_location was updated;
3. When the waiting for the third-party service ended, the metadata_location of batch 2026-02-08 was overwritten using the file of batch 2026-02-07 metadata_location.
4. This ultimately led to the data loss of batch 2026-02-08.

### Willingness to contribute

- [x] I can contribute a fix for this bug independently
- [x] I would be willing to contribute a fix for this bug with guidance from the Iceberg community
- [x] I cannot contribute a fix for this bug at this time

Start with the reported Spark workflow and the sequence of lock-survival confirmation, waiting, and metadata_location updates. Reproduce the third-party snapshot-cleanup scenario against Iceberg 1.10.1, then trace the metastore lock and metadata overwrite behavior. Done means a delayed cleanup cannot overwrite newer metadata or lose the later batch's data.

Written by the indexing model from the issue text.