Two seperate bugs cause the output of avg(latest_foreach(...., 30s)) to be non sensical if any item being queried becomes unavailable (such as if discovery no longer returns it).

1. avg appears to cast the null value from latest_foreach to 0 resulting in a low average (we have values of 100 in all items, but half of them are unavailable resulting in 50%)
2. latest_foreach doesnt support a time parameter. Documentation as to the parameters of latest_foreach is vauge but seems to indicate that foreach functions should have a time parameter "foreach_function(item filter,time period)" - if you look at the code last_foreach does not.

I'm willing to submit a patch for part 2, it looks simple enough to fix.