NA in data gives wrong results for Clim()

Hi @aho and @nperez,

the handling of NA in Clim is wrong somehow.

1st example - no NA in data. Clim() and apply() give same results:

d <- rnorm(15)
var <- array(data=d,dim=c(1,3,5,1,1,1))
names(dim(var)) <- c("dat","memb","sdate","time","lat","lon")
## 
print(Clim(var,var,memb=TRUE)$clim_exp)
print(apply(var[1,,1:5,1,1,1],1,mean))

2nd example - some random NA in data. Clim() and apply() DO NOT give same results. Instead Clim() just outputs some elements of the input data:

d <- rnorm(15)
d[c(1,2,5,6,7,14)] <- NA 
var <- array(data=d,dim=c(1,3,5,1,1,1))
names(dim(var)) <- c("dat","memb","sdate","time","lat","lon")
## 
print(Clim(var,var,memb=TRUE)$clim_exp)
print(apply(var[1,,1:5,1,1,1],1,mean,na.rm=T))
##
## result of Clim() is identical to
print(var[1,1:3,4,1,1,1])

if memb=FALSE the results are also incorrect

Thanks a lot!