Just chucking a few numbers in to see if you guys want to (and if you do how you) use them.
Assuming a direct 90 degree to surface impact, from a height equal to the orbit of the ISS (for ball park figure purposes of a realistic 'orbital descent before interception' scenario), we have a 1,057,500,000.00 kg mass operating under an 11.2mps acceleration (eve data regarding the planet in question) from a 433.4km height. This assumes the carrier to be largely intact.
A direct hit, from kinetic energy alone, will dump the equivalent of 45% of the energy that went into making the Meteor Crater in Arizona. With a barren planet such as this, it is unlikely that this will be mitigated in any way by a splash down (which would possibly actually result in more catastrophic long term ecological effects). I will leave it to better people than I to theory-craft the effects of super-heated dust/sand/iron oxide or whatever in the long term.
Put into perspective, the energy transfer is the equivalent of a directed 1.06 megatons of TNT. Higher gravity may mean slightly decreased blast radius/effects due to pressure, but as we don't know the air mix down there, I am assuming that the energy gained from increased acceleration counteracts any mitigating pressure/gravity.
If the full powergrid of the thanatos were active (unlikely after 'destruction' in orbit), it would contribute a mere 15 kilotons of TNT, potentially after the main impact event, assuming the full value of the powergrid (without skills) dumped in 1 millisecond. This is 11 x Little Boy's blast energy.
All in all, a direct kinetic hit, unless you are extremely unlucky (and have a major city right on the dot), is going to really bugger up a select area and very likely going to generate a small army of phd graduates when observation of the climate and ecological effects is done. Instead of poo pooing the idea of 3 million deaths (or a large number at least) outright, let's look at some scenarios:
1. A skidding impact could carry the main sections of the hull some way, increasing the area of effect and literally grinding everything in the skid's path to paste/rubble. Driving a 1km long, 300m wide block of super structure through the middle of a metropolitan area, even if it bled speed on a cross-country crash course, is going to severely maul the population and infrastructure. A post-landing explosion, or series of such, even on the order of 'Little Boy' level outbreaks of fusion failure, is going tto add to the carnage. Also look at infrastructural failure - gas mains and power plants react badly to impact event kinetic energy.
2. The air on the planet is toxic/thin/exposure can kill you within an hour. Any earthquake (barren worlds are tectonically inactive generally - going on their description in game - thus architecture may not account for such activity) generated by impact might breach biodomes, collapse buildings etc. Eitherway, even low levels of exposure to the impact event itself may destroy or disrupt the balance of local contained environments, causing deaths by proxy in a time frame we can consider 'immediately consequential'.
3. The break up scenario. Large chunks of heated super structure hit the planet. Combined with 2, this can lead to the containment failure of multiple biodomes.
Please bear in mind this is just a bit of fun theory crafting, not ammo for refutation/denial/argument generation from either side of any OOC debate. My intent is to provide a 'pseudoscience' framework that individuals of either side can use to 'build cases' in IC debate, should they wish. On my end, running the numbers was something to do while off work sick and already crunching figures in wolfram alpha.