Add fix and test for foldBlocks

[eyeinsky]

Creating this draft pull request in relation to the foldBlocks issue here #4624.

Due to adding genesis hashes to node config file (72e6017) cardano-testnet now does start (without adding the hashes it exceptioned out as described in the issue), but foldBlocks still doesn't receive ledger states and hangs indefinitely.

(Should I break out the 72e6017 to a separate PR so we could merge that?)

[eyeinsky]

Cardano testnet now succesfully starts and foldBlocks test (occasionally) succeeds.

I say occasionally because the test is flaky: testnet nodes do start but quite frequently in the test runs they don't start posting anything to the local chain sync protocol and the test fails with a timeout.

The good thing is that the config files now allow the testnet to start, but why the socket is not written to remains to be researched.

Another thing that might be good to pursue is that while monitoring linux processes, the cardano node's start relatively early in the test, but the local chainsync socket starts to receive updates ~20 seconds in -- is that time required, e.g spending it on reaching consensus, or could it be cut much shorter? It would greatly benefit the convenience of testing if that time could be cut shorter.

[eyeinsky]

This PR is now ready for review.

In addition to amending the node configuration file, there were two other issues that made this test flaky:

• foldBlocks doesn't fold received blocks forever but returns when latest block (or error) is reached. Sometimes this happened before any permanent blocks => permanent ledger states have been created (foldBlocks only emits permanent ledger states). We run foldBlocks in a loop to overcome this.
• The chain is not extended fast enough for the two node/one pool testnet configuration that is created by default, changing the activeSlotsCoeff from 0.2 to 0.9 fixes this.

[Jimbo4350]

Can you explain further what this test is testing? Specifically wrt the put/readMVar.

[eyeinsky]

The MVar is written to from within the handler that is passed to foldBlocks, it simply tests that a ledger state is received and the handler is called (which writes to the lock and allows the test to finish).

[Jimbo4350]

Ah right I understand. Can you include this in a comment? I'll approve after.

[eyeinsky]

Yup, will also add all fixes done here to the other PR as well (the one that's against the release/1.35 branch).

[eyeinsky]

Closing this PR in favor of this one: #4679