`extract_varcomp()` fails when variables include regular expression meta-characters.

[jepusto]

The following example returns incorrect results because the model specification includes a random effect expressed as I(session^1):

# Load packages
library(nlme)
library(scdhlm)

# Load data
data(AlberMorgan)
dat <- AlberMorgan

# Clean data
dat <- dat[,c("case", "session", "condition", "outcome")]
names(dat) <- c("case","session","phase","outcome")

dat <- preprocess_SCD(case = case, 
                      phase = phase, 
                      session = session, 
                      outcome = outcome, 
                      design = "MB", 
                      center = 5, 
                      data = dat)

# Fit the model
fit_RML <- lme(fixed = outcome ~ 1 + I(session^1) + trt, 
               random = ~ 1 + I(session^1) | case, 
               correlation = corAR1(0.01, ~ session | case), 
               data = dat,
               control = lmeControl(msMaxIter = 50, apVar = FALSE, returnObject = TRUE))
summary(fit_RML)

extract_varcomp(fit_RML)