R 将 2D 矩阵转换为 3D 数组

我有一个2D矩阵，我想将其转换为3D数组，但我没有正确地进行转换.

我的目标是让它工作，这样，无论我有多少第三维度的水平，它都会递归地工作.

以下是数据:

dfs <- (structure(c(-0.046, 0.676, -0.698, -0.047, 0.719, -0.646, -0.044, 
                             0.706, -0.667, 0.046, 0.769, -0.605, 0.047, 0.813, -0.551, 0.044, 
                             0.795, -0.578), dim = c(9L, 2L), dimnames = list(c("snatch:(Intercept)", 
                                                                                "snatch:weight", "snatch:age", "clean:(Intercept)", "clean:weight", 
                                                                                "clean:age", "total:(Intercept)", "total:weight", "total:age"
                             ), c("2.5 %", "97.5 %"))))

我要找的是这样的东西

$a
             X2.5%  X97.5%
(Intercept) -0.046   0.046
weight       0.676   0.769
age         -0.698  -0.605

$b
             X2.5%  X97.5%
(Intercept)  -0.047   0.047
weight        0.719   0.813
age          -0.646  -0.551

$c
             X2.5%  X97.5%
(Intercept)  -0.044   0.044
weight        0.706   0.795
age          -0.667  -0.578

我的 playbook 和我到目前为止所做的

rnames <- c("(Intercept)", "snatch", "age")
cnames <- colnames(dfs); cnames
lnames <- c("a", "b", "c"); lnames

array(
  data = matrix(data = type.convert(gsub("^[^:]+:\\s*", "", dfs), as.is=TRUE),
                nrow=nrow(dfs), ncol=ncol(dfs)),
  dim = c(nrow(dfs)/length(lnames), length(cnames), length(lnames)),
  dimnames = list(rows=rnames,
                  column=cnames,
                  level=lnames)) %>%
  apply(., 2, data.frame)