Add stun NAT type check for sudph

[ersonp]

Did you run make format && make check?
yes

Fixes #776

Changes:

• Added a STUN check for the sudph transport.
• Separated transports into their own modules so that they can be initialized individually and the visor will start only when at least one transport has been initialized.
• dmsg and sudph initialize separate from other modules so that the visor doesn't have to wait for them.
• Removed the unused public field from AddTransports.

How to test this PR:

1. Run ./skywire-visor skywire-config.json
2. If your visor is under a symmetric NAT you should not be able to see a sudph transport. (check the logs to see your NAT type)

[jdknives]

Looks good @ersonp . Will test and merge once we have a Stun server deployed ourselves.

[ersonp]

Few things are remaining such as:

1. Add stun servers to config
2. Put public IP and Nat type in visor summary
   Will push them soon.

[ersonp]

@jdknives @i-hate-nicknames I have turned stun into a module but corrently all it's related code is in pkg/visor/init. I think it should go under transport or transport/network the second one being from the snet refactor. Let me know if you think otherwise.

[i-hate-nicknames]

Good job overall! Also I would just remove the code that you commented out. Most likely it will be gone when snet refactoring is merged, but still a good habit

[i-hate-nicknames]

Just a small comment, otherwise looks good.

Also, now that snet refactoring got merged there is no snet package and snet.Network entity, but the changes should fit perfectly into transport manager.

[ersonp]

need to merge #842

[mrpalide]

Everything (just a little comment) is OK. Ready to merge.

[jdknives]

Looks mostly good.

[jdknives]

Please pull in develop and fix conflicts.

[ersonp]

@jdknives Currently we are not able to use sudph in integration env in skywire-services because of the STUN check. So I think we should add either a test flag in or a test field in config to turn off the STUN check.