Add support for `txt` file extensions and display hash for non text/png files

[schloerke]

Fixes #406

• Allow for diffs of *.txt files
• Execute digest::digest(raw_data, algo = "sha1") on non .(json|download|txt|png) files
• Display a message containing the hash and which files are supported.
• Add docs to app$snapshot to encourage using a .json file extension

Before PR:

With PR:

---

SHA1 spelling came from https://en.wikipedia.org/wiki/SHA-1

Benchmark of sha1, md5, and spookyhash.

a <- read_raw("pr407/tests/shinytest/test-expected-old/test")

# "small" file
bench::mark(
  md5 = digest::digest(a),
  spooky = digest::digest(a, algo = "spookyhash"),
  sha1 = digest::digest(a, algo = "sha1"),
  check = FALSE
)
#> # A tibble: 3 × 13
#>   expression      min   median `itr/sec` mem_alloc `gc/sec` n_itr  n_gc total_time result memory     time      gc
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl> <int> <dbl>   <bch:tm> <list> <list>     <list>    <list>
#> 1 md5          18.8µs   25.5µs    38118.    8.63KB    11.4   9997     3      262ms <NULL> <Rprofmem… <bench_t… <tibble …
#> 2 spooky         24µs   30.7µs    32743.    8.63KB     9.83  9997     3      305ms <NULL> <Rprofmem… <bench_t… <tibble …
#> 3 sha1         21.7µs   28.9µs    35311.    8.63KB    14.1   9996     4      283ms <NULL> <Rprofmem… <bench_t… <tibble …

# 1000x sizeof("small" file)
bench::mark(
  md5 = digest::digest(b),
  spooky = digest::digest(b, algo = "spookyhash"),
  sha1 = digest::digest(b, algo = "sha1"),
  check = FALSE
)
#> # A tibble: 3 × 13
#>   expression      min   median `itr/sec` mem_alloc `gc/sec` n_itr  n_gc total_time result memory     time      gc
#>   <bch:expr> <bch:tm> <bch:tm>     <dbl> <bch:byt>    <dbl> <int> <dbl>   <bch:tm> <list> <list>     <list>    <list>
#> 1 md5         212.8µs  262.2µs     3400.  101.19KB     6.43  1585     3      466ms <NULL> <Rprofmem… <bench_t… <tibble …
#> 2 spooky       30.1µs   36.1µs    25985.    8.63KB     7.80  9997     3      385ms <NULL> <Rprofmem… <bench_t… <tibble …
#> 3 sha1        216.7µs  236.6µs     3981.   90.73KB     8.63  1846     4      464ms <NULL> <Rprofmem… <bench_t… <tibble …

The results are still less than a millisecond, so using the familiar sha1 as the result does not need to be crypto perfect, just signal differences.

[schloerke]

All comments addressed. Merging.