The tests clearly shows that your CPU doesn't sleep - it matches results of "healthy" device if CPU is keep awake by some loop command. As a side note, it also confirms that powertop output is right, and zzztop completely lies (might be worth mentioning to zzztop maintainer). I think that you can skip any zzztop tests from now on.

Of course, it explains poor battery life (especially in standby). I guess that your active time (if used actively non-stop from charging to completely discharging) is no different than of healthy device, although in standby you loose battery power *much* more than you should (~12x more in total standby aka test1).

As it persist through reflash, I guess that hardware failure - and our wild guess of bluetooth being responsible - might be correct. Sadly, I'm not sure if you have any possibility of fixing it completely, but if yes, it would definitely upgrade your standby times (in other words = worth a shot).

I thought about who you could ask re those bluetooth messages in dmesg, but I don't have any good ideas(maybe kernel-power maintainers know something about this stuff? Just wild guess...)

/Estel