I’m not sure this is true, as the deaths from coronavirus are distinguishable based on their clinical features, as I understand it (e.g., the chest X-ray of a COVID-19 patient is readily identifiable)

Nobody is giving chest x-rays to older people here who have died, apparently of pneumonia.

I'm not a clinician, but getting to the diagnosis of pneumonia usually involves imaging, as I understand it:

https://www.aafp.org/afp/2011/0601/p1299.html

We're kinda going around in circles here.

In order to not significantly undercount the number of coronavirus deaths attributed to other causes in the US, there would need to be wide-spread testing here... which there isn't.

Imaging (by the way, you're referencing tests on live patients not dead ones) could be used as another kind of coronavirus test, though an extremely inefficient one, so probably is essentially un-used. Anyway, for pneumonia I believe we're talking about x-rays which is poor at distinguishing coronavirus (see link). Not to mention all of this is brand new. So approximately zero Radiologists are trained and using radiology in a clinical setting to diagnose coronavirus in people who have apparently died of pneumonia.