Use the open_timeout kwarg where available

[mohamedhafez]

Resolves #94

equivalent change to what @osyoyu did in ruby/net-http#224

Some justification copied over from that PR's description:

The builtin timeout in TCPSocket.open is better in several ways. First, it does not rely on a separate Ruby Thread for monitoring Timeout (which is what the timeout library internally does). Also, it is compatible with Ractors, since it does not rely on Mutexes (which is also what the timeout library does).

[osyoyu]

Looks nice! It could be worth doing something like ruby/net-http#250 to avoid nested exceptions?

[mohamedhafez]

@osyoyu done!

[mohamedhafez]

@osyoyu i also moved this rescue block here to be more a bit more closer to where its thrown and in my opinion a bit easier to understand. if you agree maybe you could do it in net-http as well:)

[mohamedhafez]

Unless the following code is also able to throw an Errno::ETIMEDOUT? Like what would happen if @open_timeout was set to like 5 hours or something, eventually the OS would time it out and throw some kind of error, right?

      Timeout.timeout(@open_timeout, Net::OpenTimeout) do
        tcp_socket(@address, @port)
      end

[mohamedhafez]

Actually that seems to throw a IO::TimeoutError, at least on Ruby 3.4, so i do think moving this rescue block does provide a bit more clarity of where the Errno::ETIMEDOUT is coming from, to whoever comes by next to edit this code

[osyoyu]

Neat, but in the net-http case, won't this lead to diverged messages between timeout impls? Failed to open TCP connection to ... is set in the re-raise in rescue

[mohamedhafez]

AH! correct in the case of net-http. There doesn't seem to be any messing with the error messages in net-smtp though so i think this little change makes sense but only for this gem

[mohamedhafez]

Made the same changes here as in net-http 0.9.0 and 0.9.1, all tests passing