test() failure not helpful for reporting exact match subset

[MichaelChirico]

test(1, c(1:10, 11:1000), c(1:10, 1000:11))
# Test 1 ran without errors but failed check that x equals y:
# > x = c(1:10, 11:1000) 
# First 6 of 1000 (type 'integer'): 
# [1] 1 2 3 4 5 6
# > y = c(1:10, 1000:11) 
# First 6 of 1000 (type 'integer'): 
# [1] 1 2 3 4 5 6
# Mean relative difference: 0.9792285

We don't get any indication of how x and y actually differ because the glimpse only shows us the first 6 entries no matter what.

For double/complex, the way to proceed is tough, because "how different is different" is very tough in general.

Still, I think we can do better than completely ignoring the issue.

[MichaelChirico]

Similar issue addressed by #7836:

test(1, NA_character_, 'abc')
# Test 1 ran without errors but failed check that x equals y:
# > x = NA_character_ 
# First 1 different of 1 (1 total, type 'character'): 
# x[1] 
#   NA 
# Non-ASCII string detected, raw representation:
# $`x[1]`
# [1] 4e 41

# > y = "abc" 
# First 1 different of 1 (1 total, type 'character'): 
#  y[1] 
# "abc" 
# 'is.NA' value mismatch: 0 in current 1 in target