Bug Fix for ymd_hms() call embedded in midnight_to_midnight

[etzkorn]

An example

fpath <- system.file("extdata", extdata_fnames[1], package = "arctools")
dat <- read_csv(fpath, skip = 10,col_types = "iiiic")
rbind(head(dat, 3), tail(dat, 3))

### passing a posix timestamp into midnight_to_midnight is bugged
acc <- dat$vectormagnitude
acc_ts <- ymd_hms(dat$timestamp)
acc2 <- midnight_to_midnight(acc, acc_ts)
plot(is.na(acc2))
plot(is.na(acc))

### the bug can be explained by the redundant call to ymd_hms within m2m
acc_ts2 <- ymd_hms(acc_ts)
plot(is.na(acc_ts))
plot(is.na(acc_ts2))