Support cells spanning multiple rows/columns

[jwodder]

Extend the Cell class from #5 to take colspan: int = 1 and rowspan: int = 1 parameters for producing cells that potentially span multiple columns and/or rows of a table.

• A multicolumn/multirow cell is passed to a Txtble at the location of the cell's top-left corner. The other cells that are "covered up by"/"merged into" a multicolumn/multirow cell must be completely absent from the data array. For example, a four-column row in which the second cell is two cells wide would be written as ["Foo", Cell("Bar", colspan=2), "Baz"] ("Baz" is here the cell in the fourth column, not the third).

  • If all the cells in a row are "covered" by other cells, the row still has to be present as an empty sequence.

• By default, the width of a multicolumn cell takes precedence over the widths of the columns it spans.

  • If the multicolumn cell is narrower than the sum of the widths of the columns it spans (plus column separators), the multicolumn cell is widened to fit.
  • If a multicolumn cell's width X is greater than the sum Y of the widths of the N columns it spans, the columns are widened to match by adding ceil((X - Y) / N) spaces to the first (X - Y) % N columns and floor((X - Y) / N) to the rest.
  • If restrict_width=True is set on a multicolumn Cell, the columns it spans will take precedence over it, width-wise.
    • If such a cell's natural width is greater than the columns it spans, its width will be reduced to match.
    • restrict_width has no effect on Cells that span only one column

• A multicolumn/multirow cell's alignment, padding, etc. is based on that of its upper-left corner.

Some basic test cases for this feature can be found on the feature/span branch.