HDDS-14114. [DiskBalancer] Fix DiskBalancer tmp directory to be under hdds/<cluster-id>/tmp

[Gargi-jais11]

What changes were proposed in this pull request?

The DiskBalancer service creates two separate tmp directories due to inconsistent path resolution:

• At startup: /data/tmp/diskBalancer (incorrect, empty, cleaned up)
• During operations: /data/hdds/CID-xxx/tmp/diskBalancer/ (correct, actually used)

[root disk1]# tree -d
.
└── hadoop-ozone
    └── datanode
        └── data
            ├── hdds
            │   └── CID-322ew-ee56-4c35-8e49-f5d7c0f35645
            │       ├── current
            │       │   ├── containerDir15
            │       │   │   ├── 8058
            │       │   │   │   ├── chunks
            │       │   │   │   └── metadata
            │       │   │   └── 8064
            │       │   │       ├── chunks
            │       │   │       └── metadata
            │       │   ├── containerDir17
            │       │   │   ├── 9008
            │       │   │   │   ├── chunks
            │       │   │   │   └── metadata
            │       │   │   └── 9068
            │       │   │       ├── chunks
            │       │   │       └── metadata
            │       │   ├── containerDir19
            │       │   │   ├── 10007
            │       │   │   │   ├── chunks
            │       │   │   │   └── metadata
            │       │   │   └── 10082
            │       │   │       ├── chunks
            │       │   │       └── metadata
            │       │   └── containerDir21
            │       │       └── 11003
            │       │           ├── chunks
            │       │           └── metadata
            │       ├── DS-aa9c8571-9e8e-4948-9c7a-b20bdd75c666
            │       │   ├── container.db
            │       │   ├── db.checkpoints
            │       │   └── db.snapshots
            │       │       └── checkpointState
            │       └── tmp
            │           ├── deleted-containers
            │           ├── diskBalancer
            │           │   └── 8048
            │           │       └── chunks
            │           └── disk-check
            └── tmp
                ├── container-copy
                └── diskBalancer 

This needs to be corrected.

Solution: Correct the initialisation logic of DiskBalancer to create diskBalancer tmp at correct place. Introduce DiskBalancer.start() to construct tmp dir.

What is the link to the Apache JIRA

https://issues.apache.org/jira/browse/HDDS-14114

How was this patch tested?

Added unit tests in TestDiskBalancerService And check using docker cluster: Before fix:

bash-5.1$ pwd
/data/hdds3/tmp/diskBalancer

After Fix:

bash-5.1$ pwd
/data/hdds3/hdds/CID-2a0edc92-45ba-4e5a-a292-08d3ed3d4809/tmp/diskBalancer