s3.clone can't run on a high concurrency env with a single bucket

During the development of MDBF-1163
which aims to replace minio (S3) (discontinued community support) with a supported S3 storage service solution (garage),

I observed that, with a single bucket (s3 db) s3.clone cannot run in a high concurrency environment where multiple MTR invocations are taking place at the same time from different builders.

High concurrency is a common scenario in Buildbot, where multiple builds
are able to run S3 tests at the same time.

I observed that the server code base includes create_database.inc
which aims to enable the support of parallel runs of the same test within a single S3 bucket.

s3.clone does not source create_database.inc .

This is the error I've got durring a concurrency test:

s3.clone                                 w3 [ retry-fail ]

        Test ended at 2026-02-04 14:07:50

CURRENT_TEST: s3.clone

mysqltest: At line 13: query 'SELECT * FROM t1 WHERE a = 99 OR b = 2' failed: <Unknown> (29): Expected object 'test/t1/index/000001' didn't exist

Where 'test/t1/index/000001' is obviously reused across MTR runs.