Make relay request timeout configurable or at least limited.

At the moment when a request to relay is made the default context with no timeout is [passed](https://github.com/flashbots/mev-boost/blob/4035cb3c8c8f9b0118a0170049203f0167c604a0/server/service.go#L361) which means the request will block until the server replies.

Later on the wait group will [block](https://github.com/flashbots/mev-boost/blob/4035cb3c8c8f9b0118a0170049203f0167c604a0/server/service.go#L465) until all the go-routines are done.

If I get it correctly, if we have a faulty relay which doesn't respond quick enough, all the other request still will be waiting.

It would be better if we could set a request timeout for each relay request.

Read server/service.go around the request creation at line 361 and the wait group at line 465 to understand how relay calls are started and awaited. Add a bounded or configurable timeout per relay request, then verify that an unresponsive relay cannot keep the overall wait blocked indefinitely.

Written by the indexing model from the issue text.