The reason for this is so that the VMFS heap size on each of the ESXi hosts is updated to the maximum value.
A workaround is to make this change on a per host basis before attempting the brownfield install. Then you won't need a reboot of the ESXi hosts.
I know Rawlinson is working on a whitepaper which will discuss this at length. I don't know when it will be published however.
HTH
Cormac