Joining on POSIXct types with mismatched timezones

[DavidArenburg]

Consider the following example

require(data.table)
wd <- structure(list(Year = c(2006L, 2006L, 2006L), day = c(361L, 361L, 
          360L), hour = c(14L, 8L, 8L), mint = c(30L, 0L, 30L), valu1 = c(0.5, 
           0.3, 0.4), Date = structure(c(1167229800, 1167206400, 1167121800
           ), class = c("POSIXct", "POSIXt"), tzone = "UTC")), .Names = c("Year", 
           "day", "hour", "mint", "valu1", "Date"), row.names = c(NA, -3L
           ), class = "data.frame") ## "UTC" time zone
setDT(wd)[, `:=`(start = Date - 1800L, end = Date + 1800L)]
#    Year day hour mint valu1                Date               start                 end
#1: 2006 361   14   30   0.5 2006-12-27 14:30:00 2006-12-27 14:00:00 2006-12-27 15:00:00
#2: 2006 361    8    0   0.3 2006-12-27 08:00:00 2006-12-27 07:30:00 2006-12-27 08:30:00
#3: 2006 360    8   30   0.4 2006-12-26 08:30:00 2006-12-26 08:00:00 2006-12-26 09:00:00

tt <- as.POSIXct(c("2006/12/27 14:23:59", "2006/12/27 16:47:59", "2006/12/27 19:12:00"), 
                         format = "%Y/%m/%d %T", tz = "Asia/Jerusalem") ## different time zone
wg <- data.table(start = tt, end = tt)
#                  start                 end
#1: 2006-12-27 14:23:59 2006-12-27 14:23:59
#2: 2006-12-27 16:47:59 2006-12-27 16:47:59
#3: 2006-12-27 19:12:00 2006-12-27 19:12:00
setkey(wg)
foverlaps(wd, wg, nomatch = 0L)[, .(wdDate = Date, valu1, WGDate = start)]
#                 wdDate valu1              WGDate
#1: 2006-12-27 14:30:00   0.5 2006-12-27 16:47:59 <--- The matched times are incorrect

See this SO question for reference.

It'd be nice to, in general, warn when joining on POSIXct types with mismatched timezones.