Fix encoding issue with docx + sub_small_vals() / sub_large_vals()

[olivroy]

Summary

Sometimes, it seems that context = "default" is not properly handled. In the case of sub_small_vals(), it led to process_text() bypassing context == "word" condition and calling htmltools::htmlEscape() which causes the issue reported in #1997. I suppose the same problem happens for sub_large_vals(), so I fixed that as well. This is probably an issue elsewhere, but let's wait for reports before engaging in that direction. I made similar fixes for grid output last year #1788

In a .qmd doc

---
format: docx
---
```{r}
tibble::tibble(x = c(0.001, 0.01, 150), 
       y = c("<", "%", ">")) -> data

table <- data %>% 
  gt() %>% 
  sub_small_vals() %>% 
  sub_large_vals(threshold = 100)
table
```

Edit: I wanted to check, and it seems we are simply removing this from the output with this change:

Edit2: I decided not to be lazy after all and tested a bit more. I added another commit to use unicode instead which changes the output to this:

Related GitHub Issues and PRs

Fixes #1997

Checklist

• I understand and agree to the Code of Conduct.
• I have listed any major changes in the NEWS.
• I have added testthat unit tests to tests/testthat for any new functionality.

[rich-iannone]

LGTM!

[rich-iannone]

Thank you @olivroy !