The recipe that claims to create an hdd image is actually just creating a partition. This can be easily confirmed by trying to loop mount the file. There is no partition table whatsoever.
Btw, I'm not fully sure of this, but it really looks like there is not way, currently, of building an efi image that has more than one partition.
My current workaround is to use bitbake to build the individual partitions and then use (as root) a template I made manually with gdisk that has the correct GTP, loop mount it and dd the partitions into it. But it's a hack and works only as long as I have one single target architecture.
I'm not sure I understood the purpose of this bug. Yes, bitbake creates non-partitioned bootable images. I thought we've agreed on direction - to to call wic from bitbake to produce partitioned images. Here is a bug about it: #7672 Please, elaborate, what do would you like to be done for this bug and how that is different from what should be done for #7672.
I see #7672 as a request for enhancement. In this case - but I might be wrong - I think the image should contain a partition table. I cannot recall ever seeing a disk image without a partition table. So to me it seemed formally incorrect to not provide a partition table. The fact that it can boot anyway might indicate otherwise. If I'm mistaken, please close the bug as invalid.
OK, now I understand that you're questioning the whole idea of the image without partition table. It's not very common, but it's definitely valid method. You can read about pros and cons of it here for example: http://unix.stackexchange.com/questions/14010/the-merits-of-a-partitionless-filesystem