serve.admin.request_log.disable_for_health has no effect: health paths missing /admin prefix in exclusion list

Describe the bug

serve.admin.request_log.disable_for_health: true does not suppress request logs for /admin/health/ready (or /admin/health/alive).

In cmd/server/handler.go, adminServer registers httprouterx.AddAdminPrefixIfNotPresentNegroni before the reqlog middleware, so reqlog sees the already-prefixed path /admin/health/ready. But the exclusion list only contains the unprefixed health paths — the /admin prefix is applied only to the Prometheus path:

logger := reqlog.
	NewMiddlewareFromLogger(d.Logger(), fmt.Sprintf("hydra/admin: %s", d.Config().IssuerURL(ctx).String()))
if cfg.RequestLog.DisableHealth {
	logger.ExcludePaths(healthx.AliveCheckPath, healthx.ReadyCheckPath, "/admin"+prometheusx.MetricsPrometheusPath)
}

n.UseFunc(httprouterx.TrimTrailingSlashNegroni)
n.UseFunc(httprouterx.NoCacheNegroni)
n.UseFunc(httprouterx.AddAdminPrefixIfNotPresentNegroni)
n.UseFunc(semconv.Middleware)
n.Use(logger)

So the exclusion never matches. The same file already builds both forms of these paths elsewhere (lines ~286-291), so the admin exclusion list looks like an oversight rather than a deliberate choice.

This is the mirror image of the equivalent Kratos bug (https://github.com/ory/kratos/issues/4605): there the prefix is added after reqlog while the exclusion list is prefixed.

Reproducing the bug

1. Run Hydra with SERVE_ADMIN_REQUEST_LOG_DISABLE_FOR_HEALTH=true.
2. curl http://<admin-host>:4445/health/ready repeatedly.
3. Every request is logged at info with msg="completed handling request" (and msg="started handling request").

Relevant log output

time=... level=info msg=completed handling request http_request=map[headers:map[user-agent:ELB-HealthChecker/2.0] host:10.0.0.1:4445 method:GET path:/admin/health/ready query:<nil> ...]

Expected behavior

Health probes against the admin listener are not logged when disable_for_health is true.

Environment

• Hydra v26.2.0, self-hosted via oryd/hydra:v26.2.0
• Config: SERVE_ADMIN_REQUEST_LOG_DISABLE_FOR_HEALTH=true

Impact

Health probes dominate our Hydra log volume (~183 MB/month of CloudWatch ingestion, essentially all /admin/health/ready from ELB-HealthChecker). Hydra also logs both started and completed lines per request, which doubles it.

Suggested fix

Add "/admin"+healthx.AliveCheckPath and "/admin"+healthx.ReadyCheckPath to the admin exclusion list.