Literally the first rant I made when I joined devRant was about this very piece of advice.
Me: "The build in testing is broken, this is how we fix it"
PM: "They haven't noticed yet, work on the feature request"
https://devrant.io/rants/225727/...

Well yeah it sucks but it makes sense if it gives you time. Also it depends on how likely a client is willing to raise an alarm because of a bug. Sometimes "cheats" like those allow you to squeeze some time in the project which you wouldn't be able to otherwise. Especially when there is a fast approaching deadline.

I find it's better to fix bugs before iterating on new features, if only because if the new code has any interaction with the original code you're going to have to fix it regardless, and as it's an indirect problem it'll probably take longer to find. I'm also not alone, it's part of the Joel test:
https://joelonsoftware.com/2000/08/...