Group and ungroup rows preserving all columns dynamically

In Power Query, I want to group my table by certain columns, add a new column based on the groups, then ungroup my table again. I can do this using the GUI. But the GUI hard-codes my column names and types. If I later want to change a column name or type, it takes a long time to update.

Can I make Power Query group and ungroup all the columns, without hard-coding the column names?
I can use Table.Schema to access the column names and the names of the types, but it's a pain to get the types themselves.

In a group you can simply remove the code between the [ ] within the group step code, and then it will operate on any columns

so that the autogenerated code

#"Grouped Rows" = Table.Group(#"Changed Type", {"a"}, {{"data", each _, type table [a=nullable number, b=nullable number, c=nullable number]}})

becomes

#"Grouped Rows" = Table.Group(#"Changed Type", {"a"}, {{"data", each _, type table }})

This is possible if you know in advance what column types are possible. You can use if [Kind]="text then type text else ... else type any to turn the type names into types. Then you have to build a record type, which uses the inconveniently-designed Type.ForRecord function.

If the previous step is called Did Previous Step, you can use the following code to group by columns "Column A" and "Column B".

=Table.Group(#"Did Previous Step", {"Column A", "Column B"}, {{"AllRows", each _,
  type table Type.ForRecord(Record.FromList(
    Table.TransformRows(
      Table.Schema(#"Did Previous Step"), 
      each [Type=if [Kind]="text" then type nullable text 
            else if [Kind]="number" then type nullable number 
            else type any, Optional=false]
    ), Table.Schema(#"Inserted Merged Column")[Name]),false)}})

Going the other way works similarly, though you should use List.Difference() twice to avoid duplicating the columns that you grouped by.