Try a slightly longer handshake timeout

[teor2345]

Motivation

The large testnet acceptance tests are failing, because sometimes Zebra only gets 1-3 good peers on testnet.

Solution

Increase the timeout to allow slightly slower peers to connect.

Review

We need to test this change before it gets reviewed.

Related Issues

#1877 Restore large testnet tests

Follow Up Work

We could also:

• increase the testnet acceptance test timeouts
• decrease the number of blocks we sync on testnet
• deploy more testnet peers
• make the security fixes that will improve DNS seeder accuracy, and re-deploy the foundation DNS seeders
• just disable the test and leave it disabled

[dconnolly]

Ah boo, still a transient failure

[teor2345]

I ran the zebrad acceptance tests --including-ignored with this PR, in a loop on my local machine, until one of the tests failed.

For the first few test runs, I got:

0 successes before a failure due to a block download timeout (from a slow peer)
13 successes before a failure due to DNS seeders providing bad peers (0 good peers)
1 success before a failure due to DNS seeders providing bad peers (3 good peers)

That's about a 20% failure rate, compared to 7% before this change:
#1877 (comment)

So this change isn't going to help, let's try the fixes and DNS seeder redeployment. See #1791 for details.