Relevel a factor column when adding a subset of current factors

[cgiachalis]

This PR is a first attempt to fix #843.

When an array has factor columns, any new writes that includes only a subset of existing factor levels will not remap these levels accurately.

Unit tests are added to cover the cases of new writes with current factors only (not growing enums).

---

Before

library(tiledb)

set_return_as_preference("data.frame")

uri <- tempfile()
df1 <- data.frame(id = 1:3, obs = factor(c("A", "B", "C")))
tiledb::fromDataFrame(df1, uri, col_index = 1, tile_domain = c(1L, 5L))
df2 <- data.frame(id = 4:5, obs = factor(c("B", "C")))
arr <- tiledb::tiledb_array(uri)
arr[] <- df2
arr[]


   id obs
1  1   A
2  2   B
3  3   C
4  4   A # <- expected B
5  5   B # <- expected C

After

   id obs
1  1   A
2  2   B
3  3   C
4  4   B
5  5   C

[mojaveazure]

Sorry for the delay in review; this looks good. I've made one suggestion below, but it's optional. Can you bump the develop version and add an entry to NEWS.md before we merge?