ArgumentOutOfRangeException: use corresponding default message for all overloads

[OwnageIsMagic]

Closes #89783

[stephentoub]

This is missing the same treatment.

[OwnageIsMagic]

This is consistent with other *Exception classes. I think this should be changed in another PR targeting all Exception derived classes
https://github.com/OwnageIsMagic/runtime/blob/3f624ff6fc8e347efde954744879c075efb1411b/src/libraries/System.Private.CoreLib/src/System/ArgumentException.cs#L44
https://github.com/OwnageIsMagic/runtime/blob/3f624ff6fc8e347efde954744879c075efb1411b/src/libraries/System.Private.CoreLib/src/System/MissingMethodException.cs#L30

[stephentoub]

But it's inconsistent with itself. There's no reason a null message to the different ctors should use a different message. Either we change them all on this type or we don't change any on this type.

[OwnageIsMagic]

What about other classes (string, Exception) ctor? Another PR?

[stephentoub]

If there are other types where this would make sense, we can use this PR to do so.

[OwnageIsMagic]

So there is no point constraining this to CoreLib.
This will affect all exception classes CoreLib (ls -1 **/*Exception.cs | wc -l # -> 100) and some in other libs
i.e.

runtime/src/libraries/System.Data.Common/src/System/Data/DataException.cs

Lines 26 to 31 in 3b21b90

	public DataException(string? s) : base(s)
	{
	HResult = HResults.Data;
	}
	public DataException(string? s, Exception? innerException) : base(s, innerException) { }

runtime/src/libraries/System.Security.Principal.Windows/src/System/Security/Principal/IdentityNotMappedException.cs

Lines 20 to 31 in 3b21b90

	public IdentityNotMappedException(string? message)
	: base(message)
	{
	}
	public IdentityNotMappedException(string? message, Exception? inner)
	: base(message, inner)
	{
	}
	internal IdentityNotMappedException(string? message, IdentityReferenceCollection? unmappedIdentities)
	: this(message)

117 classes, but not every Exception out here have custom message

---

Are you ok with proposed pattern with constructor delegation?

[stephentoub]

Are you ok with proposed pattern with constructor delegation?

I'd prefer we just add ?? SR.Whatever where each message is passed along to the base.

Delegating from ctor to ctor unnecessarily keeps all larger constructors rooted and prevents them from being trimmed away when the smaller one is being used. It's also a lot more churn.

[OwnageIsMagic]

@stephentoub test timeout on arm seems unrelated

[OwnageIsMagic]

@stephentoub ping ?

[OwnageIsMagic]

Reverted original patch and rebased on fresh main #90505