TODO (suggestions):
* use btree instead of linked list for sorting
* better parsing (instead heavy use of string functions)
* more powerful expressions in assignments/updates/conditions
* Portability (had been developed under Linux, relies on built-in
  string functions like index(), assume 1 byte char, assume enum = int)
* write more test cases to ensure functionality really works (sorting,
  beginTime/endTime
* merge multiple headers when reading from multiple files
* reconstruct header when not present but possible
* Add access variable for exceptions & processing for time lists
* real sync (slow & fast sync)
* more sanity checks, and switchable per command line option
* check & remove duplicates
* Datebk3/Datebk4 extensions (and keeping of position number in windat)
* exit on warn_message to be switchable per command line option
* csv: Outlook quoting (twice double-quote to quote double-quote, and linefeeds
  are not quoted => need to be able to read multiple lines)
* Read in Outlook export date format (which is localized, not iso-formatted)

