IOPs settings not applied when disk offering changes

### problem

Hi all.

In our PoC installation, I configured two disk offerings with different IOPs limits (1,000 and 10,000).
When I launch a test VM, the limit is applied correctly.
```xml

10000
10000

```
While the VM is running, I can change the offering, and a success message is displayed. Everything also looks successful in the management-server.log.
```
2026-05-19 16:31:47,618 DEBUG [o.a.c.f.j.i.AsyncJobManagerImpl] (API-Job-Executor-58:[ctx-e71e416f, job-1276, ctx-ee83f22a]) (logid:2c9eb69a) Complete async job-1276, jobStatus: SUCCEEDED, resultCode: 0, result: org.apache.cloudstack.api.response.VolumeResponse/volume/{"id":"d8a513c5-6241-46a9-b0d8-f969a5a245ce","name":"ROOT-61","zoneid":"57a96360-8053-487c-9fb9-85af87ca697d","zonename":"lc01","type":"ROOT","deviceid":"0","virtualmachineid":"e90057c4-5be9-492e-a78f-dfbca7507e07","templateid":"4c6bfba7-2457-466e-8767-7b3e620bb4e5","templatename":"Win2025-test2","templatedisplaytext":"Win2025-test2","vmname":"test-win02","vmdisplayname":"test-win02","vmstate":"Running","vmtype":"User","provisioningtype":"thin","size":"(50.00 GB) 53687091200","created":"2026-05-18T13:21:38+0200","state":"Ready","account":"admin","domainid":"b6a1560b-2912-11f1-af58-0050560103cb","domain":"ROOT","domainpath":"ROOT","diskIopsReadRate":"1000","diskIopsWriteRate":"1000","diskkbsread":"(1.33 MB) 1394345","diskkbswrite":"(43.25 KB) 44290","diskioread":"(52.04 KB) 53286","diskiowrite":"(3.00 KB) 3076","diskofferingid":"1b621070-6092-4d6e-91e1-987e662f2dd1","diskofferingname":"disk-1000","diskofferingdisplaytext":"disk-1000","destroyed":"false","isextractable":"false","displayvolume":"true","quiescevm":"false","supportsstoragesnapshot":"false","deleteprotection":"false","tags":[],"hasannotations":"false"}
```
However, the configuration of the VM on the KVM host remains the same.
```xml

10000
10000

```
The limit is only applied if I completely power down the VM and then power it back on.
Is there a trick to this, or something I've missed?

### versions

ACS 4.22.0
KVM/Qemu on Ubuntu 24.04

### The steps to reproduce the bug

1.
2.
3.
...

### What to do about it?

_No response_

Reproduce the issue on ACS 4.22.0 with KVM/QEMU on Ubuntu 24.04, changing a running VM between disk offerings with different IOPS limits. Compare the management-server.log result with the live VM XML before and after the change; done means the running VM's iotune values change without powering it off.

Written by the indexing model from the issue text.