It seems as if all column widths are specified as 0.0 Pandoc will calculate optimal column widths based on the length of cell contents and the available space. Is this correct?

It also seems that if a column width is an explicit non-zero float Pandoc will set the width of that column as specified so that the width of the column is the available space multiplied by the specified float, such that if the float is e.g. 0.4 the width of the column is 40% of the available width and if the sum of the specified widths is greater than 1 the table will be wider than the available width. Is this correct?

Is the above behavior, assuming I have understood it correctly, implemented in Pandoc or in LaTeX?

What happens, specifically to the columns for which a zero value is specified, if some of the specified column widths are zero and some are non-zero? I guess that the available space if any remains after subtracting the non-zero widths is distributed among the zero-width columns, or do they become truly zero-width?