Bug in time_delta_to_n_steps()

[dajmcdon]

Discovered while trying to use revision_summary() on weekly data.

When the data have weekly time_value but the version is not on the same day of the week, everything chokes.

suppressMessages(library(epiprocess))
suppressMessages(library(tidyverse))

set.seed(0)
dat <- tibble(
  time_value = seq(ymd("2020-01-01"), by = "1 week", length.out = 10),
  version = time_value + days(round(runif(10, 2, 6))),
  geo_value = rep("ca", 10),
  value = rnorm(10)
) |>
  as_epi_archive()

revision_summary(dat, value)
#> Error in `mutate()`:
#> ℹ In argument: `lag = time_minus_time_in_n_steps(version, time_value,
#>   time_type)`.
#> Caused by error in `time_delta_to_n_steps()`:
#> ! `time_delta` did not appear to contain only integerish numbers of
#>   steps between time values of time type "week"


epiprocess:::time_delta_to_n_steps(as.difftime(3, units = "days"), "week")
#> Error in `epiprocess:::time_delta_to_n_steps()`:
#> ! `time_delta` did not appear to contain only integerish numbers of
#>   steps between time values of time type "week"

^{Created on 2025-04-22 with reprex v2.1.1}

[brookslogan]

I coded this mistakenly thinking that time_value and version would be of the same "time type". Treating this as a sub-issue of #653.