handle (e.g.) `factor(x)` in RE formulas

```r
library(reformulas)
dd <- expand.grid(z=factor(1:3), temp = 1:5)
dd$y <- rnorm(nrow(dd))
form <- y ~ 1 + (1|z:factor(temp))
fr <- model.frame(subbars(form), data = dd)
mkReTrms(findbars(form), fr)
```

This fails in `mkBlist`. The problem is that model frame gets 'factor(temp)' as a column name, not 'temp', so can't evaluate "factor(temp)" in

```
eval(substitute(makeFac(fac), list(fac = ff0)), frloc)
```

Not sure how to get around this - this is a new instance of the long-standing problem that the model frame has names and values corresponding to modified/transformed variables (`log(x)`, `factor(x)`, etc.), rather than the original variable (`x`).

Reproduce the example using subbars(), model.frame(), and mkReTrms(), then inspect where mkBlist evaluates makeFac(fac) against the model frame. Compare the factor(temp) column name with the expression being evaluated. Done means the supplied random-effects formula completes successfully for transformed terms such as factor(temp).

Written by the indexing model from the issue text.