CSV files are widely used because of their simplicity and the fact that spreadsheet programs can export them or directly manipulate them. The Data Mapper can import a CSV file and create a structure definition based on the file. This requires that the first row in the file be the names of the columns which is typical.
The imported structure will have a Flat representation and specify the number of rows to skip as 1 (skipping the header information). In addition, it will generate an __osdtTerminator element at the end of the row. This is necessary to be flexible in handling a terminating newline. In some files, the last row is terminated by a newline and in others it's not. Having the optional generated __osdtTerminator element handles both of these situations. There is nothing special about the __osdtTerminator name, it's just made up so as not to conflict with the names of the columns specified in the file.
The imported structure will also have the quote handling set to optional so that any element may be quoted or not.