Does the Options attribute class really need to be sealed?

I have been using CommandLineParser for a while now (and it's changed how I build tools -- thank you!) including in a program with ~50 Verbs (and growing), and since many of the Verbs take common options, I was hoping to extend `Options` to provide common handling of them.

Simple example: if 15 Verbs take a `LogFile` option, each of which has a long name ("log-file") and a help description, I can do this now by constant strings to keep them consistent, but if I were able to do:

```
[AttributeUsage(AttributeTargets.Property)]
public class LogFileOptionAttribute : OptionAttribute
{
public LogFileOptionAttribute(bool required = false)
: base("log-file") // I don't use short option names for anything
{
HelpText = "Name of the logfile to use for blah blah blah";
Required = required;
}
}
```
and then adorn the actual option with:
```
public class BlahOptions {
...
[LogFileOption]
public string LogFileName {get ; set; }
...
}
```
I have easily a dozen of these common options. It's not a huge deal to create constant strings:
```
[Option(OPTION_LOGFILE_NAME, HelpText = OPTION_LOGFILE_HELPTEXT)]
```
but adding this level of abstraction would be more convenient.

If there's a good reason for this class to be sealed, then so be it, but mine doesn't feel like a ridiculous use case. I may well get to 100 verbs for this tool by the time I'm done.

Aucun fichier ni test n’est nommé. Commencez par localiser la déclaration C# Options/OptionAttribute et son modificateur sealed, puis examinez comment les attributs sont utilisés et couverts par les tests. C’est terminé lorsque le projet dispose d’une décision explicite et revue quant à la prise en charge des attributs d’option dérivés, avec les mises à jour correspondantes de l’API et des tests si cette décision est positive.

Rédigé par le modèle d'indexation à partir du texte de l'issue.