Small bug in AutoMl "Value prediction" scenario input

### System information

(should be irrelevant)
- Windows 10, build 2004
- .net core 3.1

### Issue

AutoMl has a scenario for regression, based on values from a txt or csv file. (named "Value prediction")
If you give it a text file, with a bunch of columns at tell it to predict one of them, there is a bug with the input.
If the column you are trying to predict, is only 1 or 0,
it will think that the column represents strings, while the column actually represents floats.
It then gives an issue that only shorts (floats) can be in the column

for example given the file:

0.1, 0.3, 1
0.6, 0.23, 1
0.8, 0.3, 0

and told to predict column 3, it will fail since it thinks the column holds strings instead of numbers