Make ens and time dimensions specific to variable being called.

[potterzot]

Change ensemble and time indices to be set according to the variable being queried rather than be fixed.

Description

ens_idx and time_idx were previously fixed by default to

ens_idx = 1:21
time_idx = 1:65

but some variables have different lengths of time dimension. Now both indices are set based on the value range of the specific variable. In addition, ens_idx and time_idx were depreciated in favor of ens and time, which allow the user to specify the values rather than the indices. This conforms to how the other dimension values (latitude and longitude) are set.

Tests are also included to check that the new values are correctly set, that indices out of bounds return an informative error, and that indices override values if both are set (i.e. if time = c(6,12) and time_idx = 1:3, the query will return times c(6,12,18) that correspond to time indices 1:3).

Related Issue

Fixes #320

[sckott]

thanks ! a few thigns:

• when I installed locally, the gefs man file was updated, can you run document again to make sure that man file is updated
• if you want people to move away from deprecated things, i think it makes sense to move to the new thing in your examples, e.g,

gefs(var, lat, lon, date = date, forecast_time = "1800", ens_idx=2:4,
      time_idx=1:8)
#> 'ens_idx' is depreciated and will be removed in future versions, please specify values (not indices) in 'ens' instead.
#> 'time_idx' is depreciated and will be removed in future versions, please specify values (not indices) in 'time' instead.

so instead change the example to use what whatever the new thing is

---

the last example, that is commented out, seems like it could be removed, especially since it fails, or should it stay?:

gefs(var, ens=1, time=1:12)
#> Error in round(lat, 0) : non-numeric argument to mathematical function

[sckott]

thanks, looks good