ensure all handlers respond with a status code, rather than rely on t…

[iwarapter]

When wrapping the server handler with a logging middleware, e.g:

type detailedResponseWriter struct {
	http.ResponseWriter
	statusCode int
}

func (dw *detailedResponseWriter) WriteHeader(code int) {
	dw.statusCode = code
	dw.ResponseWriter.WriteHeader(code)
}

func (dw *detailedResponseWriter) StatusCode() int {
	return dw.statusCode
}

func Logger(log *slog.Logger) func(next http.Handler) http.Handler {
	return func(next http.Handler) http.Handler {
		return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
			start := time.Now()
			dw := &detailedResponseWriter{ResponseWriter: w}
			next.ServeHTTP(dw, r)
			log.InfoContext(r.Context(), "request log",
				slog.String("method", r.Method),
				slog.String("request_uri", r.URL.Path),
				slog.Any("duration", time.Since(start)),
				slog.Int("status_code", dw.StatusCode()))
		})
	}
}

because most of the underlying handlers rely on the http.Server defaulting the status at the very end - it is not possible to correctly get the status for endpoints which return a status 200.

This change just ensures a consistently setting the http status from all handlers.

[q-uint]

Thanks for raising this! You're right that handlers should explicitly set status codes for observability middleware to work.

I would like to go with a slightly different approach: a ResponseWriter wrapper in ServeHTTP that ensures WriteHeader is always called explicitly, even if individual handlers forget. This way it's handled centrally rather than requiring each handler to remember.

I'll reference this PR in our commit. Appreciate the contribution!