MEV boost relationship improvements

[KaloyanTanev]

🎯 Problem to be solved

Usually when a charon cluster suffers it is from a slow MEV boost response. There are two aspects we'd like to tackle here:

1. Builder registrations failing - we have observed registrations failing, which may be the culprit of some get header calls taking 950ms (meaning it's blocking the response).
2. Get header calls taking 950ms. While it may be because of the issue mentioned above OR the MEV simply playing timing games and being on the line (waiting the maximum amount of time in order to have the biggest sample / ordering of TXs possible), it may be because of other reasons.

Both of those failures might be because we are sending multiple requests from multiple operators distributed throughout multiple geolocations.

🛠️ Proposed solution

Dig into the logs of the MEV boost of clusters that had failed proposals. Look for builder registration failures (remember we submit them once per epoch).

If there is not enough data in the logs for some builder registration failures or for late get headers, consider reaching out to the relays directly and asking them for logs on their side to see what potentially went wrong.

[OisinKyne]

Improvement on the timeouts flashbots/mev-boost#817

[KaloyanTanev]

Done