Adds validate_node_url() and refactors boot node check (#6907)

[0x7CFE]

This PR fixes #6907 by propagating error value instead of a bool flag.

IMO, full solution should return the actual reason of failure including underlying error message. However, there are some issues with it:

• Impl Display for NetworkError should handle error details
• Impl FromStr for Node and NodeEndpoint return NetworkError::AddressResolve even on incorrectly formatted addresses, like foo, enode://foo@bar and even on an empty string (Node::from_str() reports errors incorrectly #6972).

[tomusdrw]

Why introduce another method then?
Wouldn't it be better to just do is_valid_node_url().is_ok() in places that don't care about the error?

[0x7CFE]

I thought about it and it was my first intention, however, in that case, method name would conflict with it's semantics.

Also I'm a bit confused why we validate nodes by effectively doing all the work needed to construct Node just to throw it away and use the raw string value. Wouldn't it be better to return Reusult<Vec<Node>, NetworkError> from init_reserved_nodes? Anyway, I thought, that would cause substantial refactoring and is out of the scope.

For now I think it would be better to stick with the new function and refactor existing calls to is_valid_node_url.

[0x7CFE]

Removed obsolete variant is_valid_node_url().

@tomusdrw, currently validate_node_url() returns Option<NetworkError>. Maybe it would be better to return Result<(), NetworkError>? That would allow check with ? and other chaining.

Actually, we may go even further, and return Result<Node, NetworkError> but that would require re-exporting of the Node struct. Don't sure, if it's viable in this case.

[debris]

Impl Display for NetworkError should handle error details

this one is probably addressed by #6979