chore: use ed25519 keys in tests

[vasco-santos]

Needs:

• libp2p/js-libp2p-crypto#145
• libp2p/js-peer-id#132

[jacobheun]

This should now be unblocked in master.

[vasco-santos]

I will work on rebase this and get it finished

[vasco-santos]

There were some tests failing now, due to the persistent peer store, more precisely as a consequence of the following lines:

• https://github.com/libp2p/js-libp2p/blob/master/src/peer-store/persistent/index.js#L151
• https://github.com/libp2p/js-libp2p/blob/master/src/peer-store/persistent/index.js#L213

With ed25519 keys, if we create a peer from the CID, we can get its pubKey through the peerId.pubKey getter. As a consequence, we will have always pubKey. The cleanest solution here is to get from the keyBook again in: https://github.com/libp2p/js-libp2p/blob/master/src/peer-store/persistent/index.js#L213

[vasco-santos]

On another note, this PR changes all the tests that use our fixtures (the majority of the tests atm) to use ed25519 keys. For the remaining ones, the easiest solution is to change the peer-id create to use ed25519 as a breaking change release. We don't need to do it now, as well, but I think it is easier (and perhaps more correct for future changes) to do that than manually change all the tests do use it instead of RSA

[jacobheun]

With ed25519 keys, if we create a peer from the CID, we can get its pubKey through the peerId.pubKey getter. As a consequence, we will have always pubKey. The cleanest solution here is to get from the keyBook again in

For inlined public keys there's no reason to store their public key, we should avoid doing this it's just a waste of space/datastore transactions. We may want to add an easy way of checking this in peer-id. Currently only rsa keys aren't inline.

[vasco-santos]

Inline public keys check was added and tests updated accordingly

[vasco-santos]

This was added to prevent a flakey test in Firefox where sometimes the stream was still closing at this moment