With all due respect , it is the "base reason" that might be if not "false" an "assumption".
There is no doubt that there are some motherboard BIOSes that may have issues with one (or the other or all) versions of grub4dos when it is installed to the MBR (+a few hidden sectors), i.e. the boot chain is:
BIOS->grub4dos MBR (grldr.mbr)->grldr->whatever
But the normal boot chain from NT 6.x is:
BIOS->MBR->PBR->BOOTMGR
It is years that I go around saying that the good ol' Windows 2000/XP MBR (or maybe nowadays the NT 6 MBR from Windows 7) is the one that most probably the manufacturer of the BIOS have actually tested (not because they are "good" or "better", only because they are the "most common").
Most people do not experience these issues, but if you do, you can use alternate methods to boot grub4os, I don't really see the issue - when making a stick which includes a BOOTMGR - to have the "normal" NT6 boot chain:
BIOS->MBR->PBR->BOOTMGR
adding to it a BOOT.INI linked to grldr and thus have the possibility of:
BIOS->MBR->PBR->BOOTMGR->BOOT.INI->grldr
Sure, you have an added selection to make, but in a multiboot thingy you have anyway the need to make a choice.
Just for the record, in newer versions of grub4dos (but since a lot of time) you can directly use grldr instead of grldr.mbr for BOOTMGR chainloading.
In any case, second sector of the grldr.mbr has a "hole" in which it can be optionally copied the "previous" MBR (which can actually be *any* MBR, as long as it is a plain MBR one sector long), so you can do the switching from grub4dos, using (say) sector 62 to hold a copy of the first sector of the grub4dos MBR, :
root (hd0)
dd if=()0+1 of=()62+1
dd if=()1+1 of=()0+1
and then:
dd if=()62+1 of=()0+1
to return "as before", of course depending on the setup you need to provide a way to boot anyway to grub4dos or use a "dd-like" program from the booted OS to perform the same restore.
Wonko