this doesn't really look like a bug, just seems that whatever algorithm is used for the trig functions (perhaps a taylor series) involves eg. raising them to high enough powers such that precision is lost.. it would explain why the precision gets progressively worse as you start dealing with larger and larger angles.. i.e.
//echo -a $sin(360).deg $sin(3600).deg $sin(36000).deg which of course should all be 0. but, i just noticed $sin(3690).deg is 1 as expected.. doesnt really agree with that hypothesis

anyway, you could improve this by observing some trig rules like taking mod 360 since every angle is equivalent to itself mod 360. so $sin(5678).deg should be equal to $sin($calc(5678 % 360)).deg but isnt, although the latter is closer to the actual value
of course you've seen that taking mod360 isnt really enough, $sin(180).deg is still kinda wrong if only by a little
if that's still a problem there's a few ways to go from here, the first is using a little more trig and realising that, when considering these functions, every angle has a (possibly negative) equivalent acute angle. theres a few things to do here to determine the acute angle and the sign change, its not as simple as taking mod90. look up basic trigonometry for these rules
you could also of course just $round it off, 5 decimal places should do the trick. oh and 3rd option is to convert this number mod360 to radians, for some reason that appears to be more precise

//echo -a $sin(540).deg $sin($calc((540 % 360) * $pi / 180))
-0, but still numerically correct lol