如何将含多组关联列的DataFrame从宽格式转长格式（R语言）

这是我此前Stack Overflow问题的跟进提问，虽然有类似问题，但我的数据结构存在差异。

我手头有一份调查数据：受访者先选择喜欢的颜色（Q1字段为逗号分隔的颜色字符串），随后针对所选的每一种颜色回答多个后续问题（Q2、Q3），数据集的列名采用「问题描述 - 颜色」的组合格式。示例数据如下：

df <- data.frame(
    ID = 1:3,
    Q1 = c("Purple, Red, Blue", "Red, Blue, Yellow", "Red"),
    "Q2: Is it your favorite color? - Purple" = c("Average", NA, NA),
    "Q2: Is it your favorite color? - Red" = c("No", "No", "Average"),
    "Q2: Is it your favorite color? - Blue" = c("Yes", "No", NA),
    "Q2: Is it your favorite color? - Yellow" = c(NA, "Average", NA),
    "Q3: Why did you choose this color? - Purple" = c("Pretty", NA, NA),
    "Q3: Why did you choose this color? - Red" = c("Pretty", "Vibrancy", "Beautiful"),
    "Q3: Why did you choose this color? - Blue" = c("Beautiful", "Vibrancy", NA),
    "Q3: Why did you choose this color? - Yellow" = c(NA, "Pretty", NA),
    check.names = FALSE)

我期望将数据转换为每行对应单个ID的单个颜色，并关联对应的Q2、Q3值的长格式，结果如下：

我之前写的代码只能处理Q1和Q2的转换，无法扩展到Q3，代码如下：

pivot_longer(df, cols = -c(ID, Q1), values_to = "Q2") |>
    tidyr::separate_longer_delim(Q1, delim = stringr::regex(",\\s*")) |>
    filter(sub(".* ", "", name) == Q1) |>
    select(-name)

现在寻求可以同时支持Q2、Q3的宽转长解决方案。

内容的提问来源于stack exchange，提问作者Ekd