OpenXML: support row-based extraction of Excel

Suggested by Jonathan here:

  1. Ability to control the cell flow in Excel extractions. In some cases, the file content is organized by column rather than by row, as a result the TUs in the XLF don't follow the actual logical order of the file, which makes the work of linguists more confusing. Phrase TMS, for example, has an option to control the flow of the extraction (by row / by column).

This is not something I've personally encountered, so if people have example of this type of use case that they can share, please attach them.