something like this in bash would work:
echo '<workbook>'"$( unzip -pq wb.xlsx xl/worksheets/*.xml | sed -e 's/<?xml\(.*\)//g' )"'</workbook>' > wb.xml
write a file, wb.xml, containing
the concatenated contents of the unzipped files /xl/worksheets/*.xml in the zip archive wb.xlsx after removing the xml version header on each worksheet
in that order :p
I'm sure there's a java package to decompress, or "unzip", .zip files as well as a better alternative to sed for editing xml.
A word of warning about date/datetime: If xlsx is anything like xls, then date/datetime will be stored as integers/float (if I recall correctly). I believe date will be an intenger representing days with an offset of 1980-01-01 (the year). datetime is similar but with the right hand of the decimal somehow representing time.
If you're on windows you can rename .xlsx to .zip and see for yourself.