Unique ids

[dsweber2]

closes #16, ish Still could use a function that goes from id to parameters by accessing the target e.g. tar_read(forecasters) in a nice way. That variable looks like:

tar_read(forecasters) %>% print(n=20)
# A tibble: 40 × 6
# Rowwise: 
   forecaster trainer  ahead pop_scaling lags      id   
   <chr>      <chr>    <int> <lgl>       <list>    <chr>
 1 scaled_pop linreg       1 TRUE        <NULL>    ZWNiZ
 2 scaled_pop linreg       1 FALSE       <NULL>    M2RjZ
 3 scaled_pop linreg       2 TRUE        <NULL>    YzhkZ
 4 scaled_pop linreg       2 FALSE       <NULL>    ZmIwZ
 5 scaled_pop linreg       3 TRUE        <NULL>    YWM5Z
 6 scaled_pop linreg       3 FALSE       <NULL>    NzJjZ
 7 scaled_pop linreg       4 TRUE        <NULL>    ODA0M
 8 scaled_pop linreg       4 FALSE       <NULL>    MTI5Y
 9 scaled_pop quantreg     1 TRUE        <NULL>    YzQ4Y
10 scaled_pop quantreg     1 FALSE       <NULL>    N2M2Y
11 scaled_pop quantreg     2 TRUE        <NULL>    Yjk4Y
12 scaled_pop quantreg     2 FALSE       <NULL>    NzVmO
13 scaled_pop quantreg     3 TRUE        <NULL>    ZDllM
14 scaled_pop quantreg     3 FALSE       <NULL>    ZTRhY
15 scaled_pop quantreg     4 TRUE        <NULL>    MzAwM
16 scaled_pop quantreg     4 FALSE       <NULL>    ZDRkY
17 scaled_pop linreg       5 TRUE        <dbl [5]> ODY3N
18 scaled_pop linreg       5 FALSE       <dbl [5]> YzRlY
19 scaled_pop linreg       5 TRUE        <dbl [3]> ODQ5M
20 scaled_pop linreg       5 FALSE       <dbl [3]> ZTJiN
# ℹ 20 more rows
# ℹ Use `print(n = ...)` to see more rows

The target forecasters isn't actually used to generate jobs, it just records param_grid, which is bolted together from several expand_grids.

The ids are md5's that have been base64 encoded and then chopped at 5 letters; hopefully 1.07e9 is enough hashes to avoid collision.