How can I split a variable-depth, slash-delimited path_column field into separate col in KQL?

Got it 🙂

This KQL query works what you want. Notice the line with the pivot command - here you have the option to add the schema you need. The columns will then be added, even though you don't have a value for that column.
The Pivot them gives you the need to do an aggregation afterwards - so it becomes a bit messy. 

I hope this help you. Below code, can be turned into a function if you need it

let T = datatable(row:string, values:string)
[
    "row01", "col1/col2/col3",
    "row02", "col1/col2/col3/col4",
    "row03", "col1/col2",
];
T
| extend ColSplit = split(values, "/")
| mv-expand splittedColumns = ColSplit
| extend SplittedColumnsToString = tostring(splittedColumns)
| evaluate pivot(SplittedColumnsToString, count()) : (row:string, values:string, ColSplit:dynamic , NumberOfColumns:int, splittedColumns:string , col1:string, col2:string,col3:string,col4:string,col5:string,col6:string)
| project-away ColSplit, NumberOfColumns, splittedColumns
| summarize col1 = max(col1), col2 = max(col2), col3 = max(col3), col4 = max(col4), col5 = max(col5), col6 = max(col6) by row, values