Consider alternative status file formats

[schlafly]

The status file format currently contains a record for every tile in the tile file with IN_DESI != 0. We could consider moving to only recording the status of files that have been observed.

Note that the status file for afternoon planning purposes needs some information about all observed tiles and all available tiles, so even at the beginning of the survey it will need records for entire initial passes that are observable.

We've also discussed alternative forms where only observed-so-far files are recorded. These make sense, but are conceptually a bit different from the status files used in afternoon planning, which really do need to record all available tiles.