Fix a couple minor `autoplot.epi_archive` bugs

[brookslogan]

library(dplyr)
#> 
#> Attaching package: 'dplyr'
#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag
#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union
library(epiprocess)
#> Loading required package: epidatasets
#> Registered S3 method overwritten by 'tsibble':
#>   method               from 
#>   as_tibble.grouped_df dplyr
#> 
#> Attaching package: 'epiprocess'
#> The following object is masked from 'package:stats':
#> 
#>     filter

# Single version
tibble(geo_value = 1, time_value = as.Date("2020-01-01") - 1 + 1:5, version = as.Date("2020-01-01") + 6, value1 = 1:5, value2 = 1:5) %>% as_epi_archive() %>% autoplot(value1)
#> Error in seq.int(from = unclass(from), to = unclass(to), by = by): wrong sign in 'by' argument

# Multiple signals
tibble(geo_value = 1, time_value = as.Date("2020-01-01") - 1 + 1:5, version = time_value + 1, value1 = 1:5, value2 = 1:5) %>% as_epi_archive() %>% autoplot(value1, value2)
#> Error in `dplyr::mutate()`:
#> ℹ In argument: `.facets = interaction(geo_value, .response_name, sep = "
#>   / ")`.
#> Caused by error:
#> ! object '.response_name' not found

^{Created on 2025-05-14 with reprex v2.1.1}

[brookslogan]

Duplicated of #667; after submitting, GitHub claimed "This issue does not exist or has been deleted." so I re-filed, but it appeared that the original actually did get created.