divetrip rewrite

[Dirk Hohndel]

As I mentioned, on the flight last night I did a partial rewrite of the
divetrip handling. I then added code that would prevent the potential
problems that Lubimir pointed out regarding 64bit time_t values and
32bit pointers and how that could go badly wrong with the g_list
convenience functions.

All of this was just pushed into a divetrip-rewrite branch. I'm very
confident that I got this right (I had tons of time to test many
different combinations of input files and actions that in the past have
caused trouble), but I'd rather have this tested in its own branch
before it is merged into master.

Thanks

/D