[Fluxus] midi-position broken?

[Kassen]

> I have some suspicions about possible culprits. Thanks for your help,
> I'll be at this.

Well, it's turning out to be quite adventurous after all.

I fixed the issue with midi-position. That turned out to be a simple
case of bar, beat and pulse not getting initiated to zero when a new
midi-listener gets created. This may have gone unnoticed because
(re)sendng midi-start would also zero those.

This does not yet cure my midi sync for Fluxa. Something there is
quite strange and I now wonder whether Racket changed something about
the way it deals with floats. The problem seems to lie with my
function "midi-last-beat-time" which hands us the time of that beat in
a way that can easily be compared to time-now. That was working very
well, I calculated it as RTMidi got the data and handed it to Scheme,
but now it drifts between + and - 64 seconds around time-now, as
though some format conversion went wrong or a integer overflows.
Oddly midi-bar-dur and midi-beat-dur which are calculated at the same 
kind of moment and handed on in the same way seem fine.

So this isn't a issue that I now know to affect the master branch,
unless there are other similar floats that might be affected by the
same kind of thing.

Oh, well, it's a nice Sunday afternoon to play detective and learn
about types of numbers :-)

Yours,
Kas.