Add test coverage for nil Len() behavior

[veeceey]

Summary

The nil check in Len() was added in PR #95 but didn't include test coverage. This PR adds TestLenNil to verify that calling Len() on a nil *Error returns 0 instead of panicking.

Test Results

$ go test ./... -v
...
=== RUN   TestLenNil
--- PASS: TestLenNil (0.00s)
PASS
ok      github.com/hashicorp/go-multierror      0.243s

Verified manually that nil and empty multierror both return 0:

var merr *multierror.Error
fmt.Printf("Len of nil multierror: %d\n", merr.Len())  // Output: 0

merr = &multierror.Error{}
fmt.Printf("Len of empty multierror: %d\n", merr.Len())  // Output: 0

Fixes #54

[ritikrajdev]

Hey @veeceey , Thanks for the contribution. This test case looks good to me but can you please resolve the failing lint CI !

[veeceey]

Thanks for the review, @ritikrajdev! I looked into the lint failure and it's not related to my changes — the lint job uses golangci-lint-action without pinning a version, so it pulled golangci-lint v2.9.0 (latest), which passes the -buildvcs flag to go list. That flag was introduced in Go 1.18, but the repo's go.mod specifies Go 1.13, which is what the lint job uses via go-version-file: go.mod.

The lint job last passed on main on Jan 30 when an older golangci-lint version was in use. Since then, the latest version has moved to v2.9.0 and any new PR branch will hit this same failure. This is a pre-existing CI configuration issue — not caused by the test I added.

All actual test jobs (linux and windows across Go 1.13, oldstable, and stable) pass.