without a call to mpg123_scan the length returned by mpg123_length is not necessarly accurate (well, even in that case it's not guaranteed to be exact, i.e. if a file is modified while it's playing.) fixes the length reporting on some mp3 files i had around.

This changed amused to pledge "stdio rpath" early in main() and then drop down to "stdio" in ctlaction, by removing the pledge call in parse. Simplifies a bit the logic and runs more code under pledge.

see GitHub issue #1

This adds the internal management of the seeking, as well as the decoder backends bits.

makes the player process optionally send back a string describing the failure to provide a better and more coincise error message. This turns the syslog message(s): amused: unknown file type amused: failed to play, skipping /home/op/pics/phos.jpg into: amused: unknown file type; skipping /home/op/pics/phos.jpg

when the function was renamed, the error messages weren't.

this will allow to remove the `got_stop' hack in player.c

with the recent refactoring, sndio is only visible in player.c

libmpg123 is more "loud" than libmad, at least for the mp3s that I have around. Is also newer and maintained. I've heard is also faster, but amused is so simple that it doesn't make any difference.