Fix --anonymous-inbound data leak

[vtnerd]

As the title says. Hopefully get some testers to make sure nothing for broken (it is fairly basic).

[vtnerd]

After "rubber ducking" things, I decided on the following changes from the initial PR:

• Switch to std::shuffle which is not deprecated, unlike the other shuffle variants
• Set all last_seen timestamps to 0 when not in public zone.

[0xFFFC0000]

LGTM

[j-berman]

To make sure I understand the comment and its implications (which are relevant to this PR):

Let's say my node has anonymous inbound setup. I make an outgoing connection to another tor node. That outgoing node will then initiate the timed sync request to my node over the connection, and my node will respond with my node's onion address and peer ID shuffled somewhere in the peerlist. If the outgoing node then initiates connections to all nodes in the peerlist that my node sent, it will be able to map my node's peer ID to my node's onion address.

Is the above accurate?

[vtnerd]

it will be able to map my node's peer ID to my node's onion address.

The peer ID <-> onion address link happens as soon as the peerlist is sent, there's no need for another connection. Did you mean that the peer ID could be used as an identifier in other connections?

Every Tor peer id defaults to 1. Only the public zone randomizes the value.

[j-berman]

Did you mean that the peer ID could be used as an identifier in other connections?

This is what I meant. This part of the comment is what made me think it might be possible:

[The connecting peer's address] is relayed to them, iff the node has setup an inbound hidden service. The other peer will have to use the random peer_id value to link the two

Seems like it wouldn't be possible to link onion address to random peer_id thanks to the fact that every Tor peer id defaults to 1. Can you clarify the comment?

[j-berman]

Code looks good. I have a question on the comment, but code LGTM.

[jeffro256]

Is there any benefit to shuffling the entire list versus a single random insertion?

[vtnerd]

The list is already randomized (as part of anonymize = true), so the random insertion would achieve the same thing with a small speed bump. I'll update this tomorrow.

[vtnerd]

• Updated comment as per @j-berman request
• Updated random insertion as per @jeffro256 request