ideally there would be no rounding because why would there be? additionally, it doesn't specify tie breaker rule so you'll have python and js producing different results for 0.5

For some random tests, the expected length seems to be off by one?

['94366 - KzQgybHa - 51.7 43.2 56.2 36.9 46.0', '61746 - jnsGvIel - 53.4 59.2 14.0 34.4', '96362 - EDFTyzLS - 43.6 58.3 85.5', '34877 - yVLBMuTK - 59.0 25.0 60.1', '28164 - xtHvMYWi - 18.3 69.6 98.4 34.8', '29873 - DopOhJfi - 18.2 76.4 29.1 10.2', '67638 - nOXGIqOs - 55.0 89.0 82.6', '58755 - IlLRxYYZ - 94.6 79.2 40.5', '67638 - IETAxXyI - 13.1 83.4 73.2 87.4 13.3']

The attempt length (9) does not equal the answer length (8)

Yet, I'm pretty sure that the list has 9 inputs?

Python:

The requirement "Round converted temp value to an integer(!)." is technically correct, but misleadingly fails to state that validation expects NO type casting to an int if the from and to scale are the same.

For example: temp = 80.48, from_scale = 'F', to_scale = 'F' -> It should work for random inputs too: 80 should equal 80.48

I would argue that this use of mixed-type returns is bad practise (i.e. always expect int, or always expect a 2-digit float), but at the very least this requirement could be detailed in the problem statement.