is the 90% full relevant for ZFS/Zpool devices on the SSD/NVMe era?

Hi all:

With mechanical disks (aka magnetic hard disks) the 90% full limit to avoid performance problems had sense (we have 7Krpm, 10rpm and 15Krpm disks) with 512 byte sector or 4K sector size. And no matter how large was the cache speed access decreased when the inner cylinder have to be accessed.

But now; most disks are SSD even NVMe with at least a magnitude faster, even storage systems with old FC speed (4 or 8 Gbps) can have ultra fast SSD/NMVe as building layers of storage for our systems. So my question is can we allow to extend that limit to 95% or even closer to 98% without affecting the zfs/zpool performance ? Even using moderm Solaris 11.4.45 OS. Or is a underlaying issue that structurally limit performance besides storage technology?