Convert to time.Time before asserting in isNeField

[youdie006]

nefield panics on a named type whose underlying type is time.Time.

type MyTime time.Time

type Test struct {
	Start MyTime
	End   MyTime `validate:"nefield=Start"`
}
validate.Struct(Test{Start: now, End: now})
// panic: interface conversion: interface {} is validator.MyTime, not time.Time

isNeField's struct branch (baked_in.go:1072) guards on ConvertibleTo(timeType), which admits
any named type with a time.Time underlying — then asserts without converting:

if fieldType.ConvertibleTo(timeType) && currentField.Type().ConvertibleTo(timeType) {
	t := getValue(currentField).(time.Time)
	fieldTime := getValue(field).(time.Time)

getValue returns val.Interface() (validator.go:516), so the dynamic type is still MyTime
and the assertion fails.

It is the only one

Of the 28 getValue(...).(time.Time) sites in baked_in.go, 26 convert first. The two that do
not are this pair. isEqField:1402, isGteField:2326, isGtField:2370, isLteField:2554,
isLtField:2598 and isNeCrossStructField:1308 — the same not-equal semantics in the sibling
family — all do:

t := getValue(currentField.Convert(timeType)).(time.Time)

Same input through four of them, before the change:

eqfield   -> ok
ltfield   -> ok
necsfield -> ok
nefield   -> panic: interface conversion: interface {} is validator.MyTime, not time.Time

Change

Two lines, adding the Convert(timeType) the siblings already have. Convert is a no-op for a
plain time.Time, so nothing on the currently covered paths changes, and the v10 API is untouched.

Test added next to the other nefield tests. Verified red/green by reverting only baked_in.go
and keeping the test — it panics with interface conversion: interface {} is validator.namedTime.
It also pins the semantics rather than just the absence of a panic: equal named-time values must
fail the tag, different ones must pass.

go test -cover -race ./... passes across every package. gofmt -l reports doc.go and
translations/ko/ko_test.go both before and after, so this branch does not change it.

validator_test.go:5317-5414 covers nefield against time.Time, *time.Time and
time.Duration, which is why this shape was never exercised.

---

Disclosure: prepared with AI assistance; I verified the reproduction, the red/green runs and the
race suite myself.

[youdie006]

The lint job is red on master too — 51 exhaustruct_v5 findings in benchmarks_test.go and
validator_test.go, none of them on the lines this PR adds (my test is validator_test.go
1943-1962; the flagged lines are 356…903 and 2207…10033). All six test jobs pass on Go 1.25 and
1.26 across Linux, macOS and Windows.

[nodivbyzero]

The lint job is red on master too — 51 exhaustruct_v5 findings in benchmarks_test.go and validator_test.go, none of them on the lines this PR adds (my test is validator_test.go 1943-1962; the flagged lines are 356…903 and 2207…10033). All six test jobs pass on Go 1.25 and 1.26 across Linux, macOS and Windows.

I opened a PR to fix the linter issues: #1618

[nodivbyzero]

Thank you for your contribution!
Please rebase your PR and I'll merge your changes once that's done.