Log Level settings don't work as expected

[jeff-mccoy]

We probably need to verify the E2E for this, but the way the log levels are currently configured shouldn't work properly (and doesn't for me). Issue is we are using logrus.New() and then setting the log level vs just logrus.SetLevel(). Recommend we just push that to the root.cmd such as below:

var rootCmd = &cobra.Command{
	PersistentPreRun: func(cmd *cobra.Command, args []string) {
		setLogLevel(zarfLogLevel)
	},
...

func setLogLevel(logLevel string) {
	switch logLevel {
	case "debug":
		logrus.SetLevel(logrus.DebugLevel)
	case "info":
		logrus.SetLevel(logrus.InfoLevel)
	case "warn":
		logrus.SetLevel(logrus.WarnLevel)
	case "error":
		logrus.SetLevel(logrus.ErrorLevel)
	case "fatal":
		logrus.SetLevel(logrus.FatalLevel)
	case "panic":
		logrus.SetLevel(logrus.PanicLevel)
	default:
		logrus.Fatal("Unrecongized log level entry: %s", logLevel)
	}
}