fix(validators): distinguish a missing package from a missing version on PyPI/NPM (#1411)

Fixes #553.

The PyPI and npm validators fetch the version-specific metadata endpoint
(`/pypi/{name}/{version}/json` and the npm equivalent) and report any
non-200 as `<pkg> not found`. That endpoint 404s for two different
reasons though: the package doesn't exist, or it exists and only the
*version* is missing (e.g. a release that hasn't propagated yet). So a
valid package reads as gone:

```diff
- PyPI package 'requests' not found (status: 404)
+ PyPI package 'requests' exists, but version '99.99.99' was not found (status: 404). A newly published release can take a moment to appear on PyPI. Wait and retry, or publish version '99.99.99' before registering it
```

On a version 404 it probes the package-level endpoint
(`/pypi/{name}/json`, `/{name}`) with a `HEAD` to tell the two apart,
and reports 429/5xx as transient rather than "not found". `HEAD` so it
reads a status without pulling the whole packument. The probe carries
its own 3s deadline: it only refines the error message, so a hung probe
must not stretch the validator past the ~10s-per-registry budget the
publish path assumes. Same probe-and-classify shape as the existing
cargo validator.

The fetch is split into `validatePyPIPackage` / `validateNPMPackage`
behind `export_test.go` so the branches are testable with `httptest`,
the same seam cargo uses.

### Testing
- [x] hermetic `httptest` for the status matrix (version-missing,
package-missing, 5xx, 429, inconclusive probe, positive path, scoped
npm); the mocks pin the expected method per endpoint (`GET` fetch,
`HEAD` probe)
- [x] hermetic deadline test: a hung probe is cut off at ~3s and
reported as inconclusive, not "not found"
- [x] existing live package tests still pass, now also hitting the HEAD
probe; on a live 429/5xx they now `t.Skip` as inconclusive instead of
flaking (deliberate: CI runs these against the real registries with no
short-mode gating)
- [x] `gofmt` / `go vet` / `golangci-lint` clean

Out of scope: auto-retry for the propagation race, and SSRF
redirect-pinning parity for the pypi/npm clients (pre-existing, shared
with nuget).

internal/validators/registries/npm.go

@@ -52,6 +52,14 @@ func ValidateNPM(ctx context.Context, pkg model.Package, serverName string) erro
 			pkg.RegistryBaseURL, model.RegistryTypeNPM, model.RegistryURLNPM)
 	}
 
+	return validateNPMPackage(ctx, pkg, serverName)
+}
+
+// validateNPMPackage performs the version-metadata fetch and the mcpName check.
+// It is split out from ValidateNPM so that httptest-based tests can drive the HTTP
+// pipeline against a mock server (exposed via export_test.go), bypassing the
+// exact-baseURL guard that ValidateNPM enforces for callers.
+func validateNPMPackage(ctx context.Context, pkg model.Package, serverName string) error {
 	client := &http.Client{Timeout: 10 * time.Second}
 
 	requestURL := pkg.RegistryBaseURL + "/" + url.PathEscape(pkg.Identifier) + "/" + url.PathEscape(pkg.Version)
@@ -60,7 +68,7 @@ func ValidateNPM(ctx context.Context, pkg model.Package, serverName string) erro
 		return fmt.Errorf("failed to create request: %w", err)
 	}
 
-	req.Header.Set("User-Agent", "MCP-Registry-Validator/1.0")
+	req.Header.Set("User-Agent", userAgent)
 	req.Header.Set("Accept", "application/json")
 
 	resp, err := client.Do(req)
@@ -70,7 +78,7 @@ func ValidateNPM(ctx context.Context, pkg model.Package, serverName string) erro
 	defer resp.Body.Close()
 
 	if resp.StatusCode != http.StatusOK {
-		return fmt.Errorf("NPM package '%s' not found (status: %d)", pkg.Identifier, resp.StatusCode)
+		return npmFetchStatusError(ctx, client, pkg, resp.StatusCode)
 	}
 
 	var npmResp NPMPackageResponse
@@ -88,3 +96,90 @@ func ValidateNPM(ctx context.Context, pkg model.Package, serverName string) erro
 
 	return nil
 }
+
+// npmPackageState is the outcome of probing the package-level NPM metadata
+// endpoint, used to disambiguate a 404 from the version-specific endpoint.
+type npmPackageState int
+
+const (
+	// npmPackageUnknown: the probe returned a status we can't classify.
+	npmPackageUnknown npmPackageState = iota
+	// npmPackageExists: the package exists (200) but the requested version does not.
+	npmPackageExists
+	// npmPackageMissing: the package itself does not exist (404).
+	npmPackageMissing
+	// npmPackageTransient: the probe failed for a retryable reason (network error,
+	// 429, or 5xx). Existence is undetermined and the caller should not report
+	// "not found".
+	npmPackageTransient
+)
+
+// npmFetchStatusError maps a non-200 status from the version-specific metadata
+// endpoint to a caller-actionable error. A 404 is delegated to npmVersion404Error
+// for disambiguation. 429/5xx are transient, not "not found".
+func npmFetchStatusError(ctx context.Context, client *http.Client, pkg model.Package, status int) error {
+	switch {
+	case status == http.StatusNotFound:
+		return npmVersion404Error(ctx, client, pkg)
+	case status == http.StatusTooManyRequests:
+		return fmt.Errorf("NPM rate-limited the metadata request for package '%s' (status: 429). Likely transient, retry later", pkg.Identifier)
+	case status >= 500 && status < 600:
+		return fmt.Errorf("NPM upstream error fetching metadata for package '%s' (status: %d). Likely transient, retry later", pkg.Identifier, status)
+	default:
+		return fmt.Errorf("NPM package '%s' metadata fetch failed (status: %d)", pkg.Identifier, status)
+	}
+}
+
+// npmVersion404Error disambiguates a 404 from the version-specific endpoint: a
+// genuinely-missing package versus a package that exists but whose requested
+// version is absent (commonly because a freshly published release has not yet
+// propagated). It probes the package-level endpoint so the publisher gets an
+// actionable message rather than a blanket "not found".
+func npmVersion404Error(ctx context.Context, client *http.Client, pkg model.Package) error {
+	switch probeNPMPackage(ctx, client, pkg.RegistryBaseURL, pkg.Identifier) {
+	case npmPackageExists:
+		return fmt.Errorf("NPM package '%s' exists, but version '%s' was not found (status: 404). A newly published release can take a moment to appear on the registry. Wait and retry, or publish version '%s' before registering it", pkg.Identifier, pkg.Version, pkg.Version)
+	case npmPackageMissing:
+		return fmt.Errorf("NPM package '%s' not found (status: 404)", pkg.Identifier)
+	case npmPackageTransient:
+		return fmt.Errorf("NPM could not confirm package '%s' version '%s' (version status: 404, package check inconclusive). Likely transient, retry later", pkg.Identifier, pkg.Version)
+	case npmPackageUnknown:
+		// Probe returned an unclassifiable status, so fall through to the
+		// best-effort message below.
+	}
+	return fmt.Errorf("NPM package '%s' version '%s' not found (status: 404)", pkg.Identifier, pkg.Version)
+}
+
+// probeNPMPackage checks whether a package exists on the NPM registry regardless
+// of version, with a HEAD request to the package-level endpoint (/{name}). Only
+// the status code is used, so HEAD avoids downloading the (large) packument.
+func probeNPMPackage(ctx context.Context, client *http.Client, baseURL, identifier string) npmPackageState {
+	// The probe only refines the 404 error message, so it must not extend the
+	// validator's worst case by another full client timeout.
+	ctx, cancel := context.WithTimeout(ctx, 3*time.Second)
+	defer cancel()
+
+	probeURL := baseURL + "/" + url.PathEscape(identifier)
+	req, err := http.NewRequestWithContext(ctx, http.MethodHead, probeURL, nil)
+	if err != nil {
+		return npmPackageUnknown
+	}
+	req.Header.Set("User-Agent", userAgent)
+
+	resp, err := client.Do(req)
+	if err != nil {
+		return npmPackageTransient
+	}
+	defer resp.Body.Close()
+
+	switch {
+	case resp.StatusCode == http.StatusOK:
+		return npmPackageExists
+	case resp.StatusCode == http.StatusNotFound:
+		return npmPackageMissing
+	case resp.StatusCode == http.StatusTooManyRequests, resp.StatusCode >= 500 && resp.StatusCode < 600:
+		return npmPackageTransient
+	default:
+		return npmPackageUnknown
+	}
+}