Cardano Updates

I don't trust it. Compare these two runs:
1. https://github.com/input-output-hk/cardano-haskell-packages/actions/runs/4532746158/jobs/7984714895
2. https://github.com/input-output-hk/cardano-haskell-packages/actions/runs/4516172089/jobs/7954255360

The second is on the parent commit of the first, the commit itself is a
no-op, and yet the derivations
they build are not the same! Doing the same locally, I get a) a
different derivation, but b) the same derivation for both.

I conclude that what the CI is doing is questionable, the cache seems
like the most likely source of pollution.

I suspect whatever is causing this was also responsible for the no-op PR
spending a lot of time building tons of stuff.

Co-authored-by: Sebastien Guillemot <[email protected]>

5018: Use ouroboros-network-0.3.0.2 and ouroboros-network-framework-0.2.0.1 r=dermetfan a=coot



Co-authored-by: Marcin Szamotulski <[email protected]>
Co-authored-by: Samuel Leathers <[email protected]>

Waiting for a day, at a pace of one block every 20 seconds would
generate 4320 blocks which takes forever to be created and
processes, even in IO sim.

We don't really observe the issue now because we cheat in the
way we generate the block.

This commit helps preparing for a more realistic block generation.

Everyone collects after observing the last commit

That way, we try to get closer to an actual block production pattern.