Add tests for weekly spaced data

The motivating case is this outdated snippet from `step_epi_slide`, yoinked from https://github.com/cmu-delphi/epipredict/pull/387
```
tt_week <- seq(as.Date("2022-01-01"), by = "1 week", length.out = 20)
edf_weekly <- data.frame(
time_value = c(tt_week, tt_week),
geo_value = rep(c("ca", "ny"), each = 20L),
value = c(2:21, 3:22)
) %>%
as_epi_df()

test_that("epi_slide works on weekly data with one of before/ahead set", {
expect_no_error(
baked <- epi_recipe(edf_weekly) %>%
step_epi_slide(value, .f = "mean", before = as.difftime(3, units = "weeks")) %>%
prep(edf_weekly) %>%
bake(new_data = NULL)
)
attributes(baked)$metadata$as_of <- as.Date("1999-09-09")
expect_snapshot(baked)
})
```

Start by locating the existing tests for step_epi_slide and the epi_recipe pipeline. Add the weekly-data case shown in the issue, covering a slide with before set and ahead unset, then run the relevant test file and update the snapshot. Done means the weekly example runs without error and its baked output is recorded.

Written by the indexing model from the issue text.