Sure, if a mutation that makes it give deadly miocarditis 50% of the cases pops up, that strain would have a clear evolutionary disadvantage ...but if a strain that gives 2x worse pneumonia (and maybe a 2x death rate, that overall is still low) with 4x more coughing evolve, it will have an evolutionary advantage, more caughing => more virus in the air etc.
So you absolutely can have viruses evolving towards increased lethality, as long as people infected with the new strain spread the virus more and don't die too soon... you won't get to smth. evolving its way to 80% lethality, but it can very well evolve from <1% to ...50% (scary!).
The whole optimism about patterns of viral evolution is 100% wrong, unless you're thinking at very large timescales, and in environments where humans have not added so many accelerator factors!