That matches what HPE documents for AOS-CX. A standalone 8360 has no ISSU path, so the control and data plane both bounce on every image change. The near-hitless option on that platform is VSX: vsx update-software upgrades the secondary first, then the primary, while traffic stays on the surviving member. Without a pair you are stuck with a cold reload.
What still helps on a standalone box is staging onto the unused bank ahead of time (copy tftp://... secondary, then boot system secondary and check with show images) so the maintenance window is just the reboot, not the download. Same class of downtime as a non-ISSU Cisco that only cold-upgrades.