Harden Streamgraph data loading by removing fragile covid_month.csv remote dependency

[Copilot]

The render failed at DataOverTime/Streamgraph.qmd while reading covid_month.csv from COS (cannot open the connection), which aborted the full babelquarto::render_website() job. The failure originated from a non-essential second remote read in the tutorial data-prep chunk.

• Root-cause scope

  • Failure point was specific to covid_month.csv fetch in Streamgraph EN/ZH tutorials.
  • covid_all.csv read remained sufficient to derive equivalent monthly aggregates used for demonstration.

• Change made (EN + ZH parity)

  • Updated:
    • DataOverTime/Streamgraph.qmd
    • DataOverTime/Streamgraph.zh.qmd
  • Replaced direct readr::read_csv(".../covid_month.csv") with a deterministic derivation from covid_all:
    • extract month from time
    • group by location, month
    • summarize count

• Behavioral impact

  • Removes a flaky network dependency from the load-data chunk.
  • Keeps tutorial semantics intact by producing covid_month from already-loaded source data.

covid_month <- covid_all %>%
  dplyr::mutate(month = as.integer(format(as.Date(time), "%m"))) %>%
  dplyr::group_by(location, month) %>%
  dplyr::summarise(count = sum(count), .groups = "drop")

[ShixiangWang]

偶发性的数据不可读？