Handle single prediction case for multinomial glmnet

[juliasilge]

This PR fixes the multinomial glmnet prediction for the single observation case that, it turns out, was half-fixed in #395. I did not realize it was only half-fixed because of how those post operations are not really loaded as functions and the test I wrote in extratests didn't start getting run until just now when we fixed the version number in the DESCRIPTION file.

So now this fixes it for real:

library(parsnip)

data("hpc_data", package = "modeldata")
hpc <- hpc_data[, c(2:5, 8)]

xy_fit <- fit_xy(
  multinom_reg(penalty = 0.1) %>% set_engine("glmnet"),
  x = hpc[, 1:4],
  y = hpc$class
)

predict(xy_fit, hpc[3, 1:4], type = "class")
#> # A tibble: 1 x 1
#>   .pred_class
#>   <fct>      
#> 1 VF
predict(xy_fit, hpc[3, 1:4], type = "prob")
#> # A tibble: 1 x 4
#>   .pred_VF .pred_F .pred_M .pred_L
#>      <dbl>   <dbl>   <dbl>   <dbl>
#> 1    0.531   0.298   0.114  0.0574

predict(xy_fit, hpc[10:11, 1:4], type = "class")
#> # A tibble: 2 x 1
#>   .pred_class
#>   <fct>      
#> 1 VF         
#> 2 VF
predict(xy_fit, hpc[10:11, 1:4], type = "prob")
#> # A tibble: 2 x 4
#>   .pred_VF .pred_F .pred_M .pred_L
#>      <dbl>   <dbl>   <dbl>   <dbl>
#> 1    0.524   0.302   0.115  0.0581
#> 2    0.524   0.302   0.115  0.0581

^{Created on 2021-02-09 by the reprex package (v1.0.0)}

[topepo]

Thanks!

[github-actions]

This pull request has been automatically locked. If you believe you have found a related problem, please file a new issue (with a reprex: https://reprex.tidyverse.org) and link to this issue.