I think we've found a possible workaround for the moment, to avoid the crashes (we are in contact with HPE with an open case, in case it's a hardware or firmware issue). We are not seeing the crashes after changing de discard option at the pool level to nopassdown. As far as we know it shouldn't have a negative effect, since in theory the controller and disks manage their own TRIM and garbage collection. If anyone knows more about that change, any comment or suggestion would be appreciated.

Thank you for the information about the workaround.
This helped someone else as well already.

Booted Ubuntu Server 26.04 LTS from ISO and could not reproduce the crash with the same fio workloads

Do you remember the exact kernel version?
That would help us narrow down any changes our kernels have, that the Ubuntu one did not.
Since our kernels are based on the Ubuntu kernels, those differences should be minor.