IAT数据处理求助：按指定规则筛选合格被试（R语言）

我正在开展IAT（内隐联想测验）研究，需遵循Nosek等人的标准流程与计分算法筛选被试。参考脚本已生成P_300（<300ms反应占比）、P_error（错误率）字段，test字段标记练习块(1)与关键块(2)。已完成以下R语言初步数据处理代码（注：当前代码移除了test字段，后续筛选需重新关联该字段以区分练习块与关键块）：

library(tidyverse)
# libraries ----
library(tidyverse)

# load raw data ----
iat_raw <- read.delim("data/iat_raw.xlsx",
  header = TRUE,
  sep = "\t",
  fill = TRUE,
  colClasses = rep("numeric", 7)
)

# flag and summarise data ----
iat_1 <- iat_raw %>%
  select(-c(order, test)) %>%
  mutate(
    lat300 = if_else(latency < 300, 1, 0),
    lat3k = if_else(latency > 3000, 1, 0),
    lat400 = if_else(latency < 400, 1, 0),
    lat10k = if_else(latency > 10000, 1, 0)
  ) %>%
  group_by(id, blocknum) %>%
  summarise(
    # mean standraddeviation for block latencies including all trials 
    M_latency = round(mean(latency),0),
    SD_latency = round(sd(latency),3),
    M_correct = round(mean(correct),3),
    SD_correct = round(sd(correct),3),
    # trial counts
    N_trials = n(),
    N_correct = sum(correct),
    N_error = n() - sum(correct),
    N_300 = sum(lat300),
    N_3k = sum(lat3k),
    N_400 = sum(lat400),
    N_10k = sum(lat10k),
    # percentages of trial counts
    # P_correct = round(N_correct / n(), 3), # the same like M_correct
    P_error = round(N_error / n(), 3),
    P_300 = round(N_300 / n(), 3),
    P_3k = round(N_3k / n(), 3),
    P_400 = round(N_400 / n(), 3),
    P_10k = round(N_300 / n(), 3)
  ) %>%
  ungroup() %>%
  arrange()

现需实现以下7条不合格被试排除规则（需排除整个不合格被试而非仅标记块），重点解决「任意块」及「跨块汇总」的筛选逻辑：

• 总测验中<300ms反应占比超10%；
• 任意关键块中<300ms反应占比超25%；
• 任意练习块中<300ms反应占比超35%；
• 所有关键块总错误率超30%；
• 任意关键块错误率超40%；
• 所有练习块总错误率超40%；
• 任意练习块错误率超50%。

请协助完成后续代码编写。

内容的提问来源于stack exchange，提问作者marsellus w.