Inefficient, yes. But no input validation is not a knock against this solution.

As per the details of this Kata:
"All the test strings would contain valid Morse code, so you may skip checking for errors and exceptions."

Hmm very cool 3 liner, but little bit slow.

Extremely unefficient solution with so many nested loops and totally unnecessary toList conversion.
Also, to me it doesn't look like so expressive.
Last but not least, there is no input validation at all.

Far away to be considered as 'Best practice'.

There's more to a Best Practice than asymptotic performance, e.g. readability and expressiveness.

Out of curiosity, what do you think the asymptotic performance of this solution is?

This has nested loops and can not be considered as a Best Practice. I have implemented this in O(n) time order.