label_bquote turns numeric values into factors (?) when as.table = FALSE in facet_grid

I found a problem with label_bquote. I was trying to use an expression show the value of a parameter in the facet labels and I found that when I used as.table = FALSE the values of the rows were instead turned into integers. I am guessing this is because the values were first turned into factors and then faceted, but I am not sure.

Here is the code to reproduce the bug:

library(tidyverse)

expand_grid(a = c(0.15, 0.85),
            b = c(0.001, 0.0025, 0.005), 
            x = rnorm(10), 
            y = rnorm(10)) %>% 
ggplot(., aes(x,y)) + 
  geom_point() + 
  facet_grid(a ~ b, 
             as.table = F, # <- works when this is TRUE
             labeller = label_bquote(rows = alpha==.(a), 
                                     cols = beta==.(b)))

Start by running the reproducible R example with ggplot2, facet_grid(), as.table = FALSE, and label_bquote(). Trace how row values are transformed while facet labels are built, then check the related labelling and faceting tests. Done means numeric row values such as 0.15 and 0.85 remain correctly represented in the labels when as.table is FALSE.