Efficient way to conditionally transform nested tables with different structures.

After a couple of days messing around with this on and off before posting my question I have found a method that doesn't involve adding and removing columns thanks to expanding on this post https://stackoverflow.com/questions/31548135/power-query-transform-a-column-based-on-another-column 

Code-wise I think METHOD_2 is a little easier to grasp, but the purist in me likes METHOD_3 not adding extra columns to clean up.

Writing up the whole question with simpler examples than my real-world PDF mess gave me some insight and clarity on how all this works. 

//////////////////////////////////////////////////////////////////////
// METHOD_3 : Insert a new column and transform conditionally
//////////////////////////////////////////////////////////////////////

METHOD_3 = Table.FromRecords(Table.TransformRows( MAIN_TABLE, (r) =>  Record.TransformFields( r, {
{"Data", each 
   if r[Id] = "Table004" then 
      Table.Skip(_, each [Column1] <> "Date")

   else if r[Id] = "Table006" then
      [  FirstRow       = Table.FirstN( _, 1 ),
         ReplaceRow1LF  = Table.ReplaceValue( FirstRow , "#(lf)", " ", Replacer.ReplaceText, Table.ColumnNames(_) ),
         RecombinedRows = ReplaceRow1LF & Table.Skip( _, 1), 
         PromoteHeaders = Table.PromoteHeaders( RecombinedRows ) 
      ][PromoteHeaders]

   else _
}}))),