simplify tests with cmp.Diff

[alexandear]

This PR refactors tests. These code blocks

if !cmp.Equal(want, got) {
	t.Fatal(cmp.Diff(want, got))
}

if !bytes.Equal(want, got) {
	t.Fatal(cmp.Diff(want, got))
}

if want != got {
	t.Fatal(cmp.Diff(want, got))
}

changed to the following:

if diff := cmp.Diff(want, got); diff != "" {
	t.Fatal(diff)
}

[rhysd]

I intentionally avoid cmp.Diff because it requires extra performance cost to generate diff comparing to cmp.Equal.

[alexandear]

I intentionally avoid cmp.Diff because it requires extra performance cost to generate diff comparing to cmp.Equal.

Could you provide more details on the performance difference between using cmp.Diff and cmp.Equal? Specifically, how much additional time does cmp.Diff take compared to cmp.Equal in tests or benchmarks?

Tests running time on my machine shows there is no significant difference.

Before:

❯ time go test -shuffle=on -count=1 ./...
?       github.com/rhysd/actionlint/cmd/actionlint      [no test files]
ok      github.com/rhysd/actionlint     2.128s
ok      github.com/rhysd/actionlint/scripts/check-checks        1.216s
ok      github.com/rhysd/actionlint/scripts/generate-availability       1.350s
ok      github.com/rhysd/actionlint/scripts/generate-popular-actions    2.501s
ok      github.com/rhysd/actionlint/scripts/generate-webhook-events     1.776s
go test -shuffle=on -count=1 ./...  2.50s user 1.67s system 139% cpu 2.999 total

After:

❯ time go test -shuffle=on -count=1 ./...
?       github.com/rhysd/actionlint/cmd/actionlint      [no test files]
ok      github.com/rhysd/actionlint     1.946s
ok      github.com/rhysd/actionlint/scripts/check-checks        1.183s
ok      github.com/rhysd/actionlint/scripts/generate-availability       0.814s
ok      github.com/rhysd/actionlint/scripts/generate-popular-actions    1.653s
ok      github.com/rhysd/actionlint/scripts/generate-webhook-events     1.673s
go test -shuffle=on -count=1 ./...  2.50s user 1.63s system 175% cpu 2.346 total

[rhysd]

Today I compared the performance and actually I couldn't find any performance differences between the APIs. I remember that there was some performance issue (due to a lot of allocations) in the past on my local but now it no longer exists.

[rhysd]

Merged. Thanks!