The primary change in the df output is the addition of the initramfs. This was there in squeeze too, but hidden since the mtab was actually missing this information. From this point of view, now we use /proc/mounts, what you're getting is the full set of mounts.
There's a second issue, which is that the used/available values for the initramfs are incorrect due to it not being possible to stat the filesystem. This is really a coreutils df bug; it should probably leave blank or zero or undefined in some way.
Regarding the UUID issue: I don't see this on any of my systems. On some, I have LVM root, and there I have the proper LV path. On the [unstable] system I'm writing this, I have a Btrfs root, and even though I have a UUID in my fstab, df is reporting the actual device node (sda4). It's possible that the issue is already fixed in unstable, or it doesn't affect Btrfs.
As for why it wasn't fixed in wheezy: once we were frozen, this was not considered a release-critical issue. At most, it's a cosmetic annoyance. It has zero impact upon the correct functioning of the system. Eliminating static /etc/mtab in favour of a /proc/mounts symlink fixed a number of rather more serious bugs, so the tradeoff was well worth it.