formatspecs, making for much cleaner code. Reformatted everything.
This actually happened in multiple steps but then I corrupted my local
repository and I had to recover from the working tree.
This prevents a crash when more tracks were recorded than fit in an
scp image. When I first encountered a problem doing an 84-track operation,
this led to a segfault later on since memory was written past the end
of _fileheader, and (I think) stomped on a pointer.
There were a couple of problems that needed to be fixed:
- Fluxmaps can be produced with zero index pulses by encoders.
This is somewhat ambiguous but it's assumed that this means there
is exactly one revolution.
- The SCP unindexed mode wasn't properly supported, as it should
create equal-length revolutions not aligned to indexes. However,
fluxmaps don't contain any information on rotation speed. Therefore,
drop the ability to create unindexed SCPs and slightly change the
meaning of the option to not include pre-index data.
- Real flux reads can potentially produce duplicate index pulses.
Avoid creating zero-length revs.
- Fix off-by-one error for rev count.
