CSVDataSet might produce wrong results in SHARE_GROUP or SHARE_THREAD modes as it relies on System.identityHashCode uniqueness for ThreadGroup and Thread

### Expected behavior

`CSVDataSet` should not rely on `System.identityHashCode` uniqueness as it is never guaranteed.

I think `CSVDataSet` should leverage smart cloning like in https://github.com/apache/jmeter/pull/5916, so it would not require accessing `System.identityHashCode(threadGroup)` every time.
As a bonus, the alias computation could be done just once, not for every execution.

See https://github.com/apache/jmeter/pull/693

### Actual behavior

_No response_

### Steps to reproduce the problem

Add a test that configures `CSVDataSet` with `SHARE_GROUP` or `SHARE_THREAD`. Use `-XX:hashCode=2` to trigger OpenJDK Hotspot to produce duplicate values for `System.identityHashCode`.

### JMeter Version

5.5

### Java Version

_No response_

### OS Version

_No response_