eth/catalyst: move block commit into its own go-routine to avoid deadlock

[jwasinger]

Closes #29475 . The provided test-case captures the issue and fails without the fix.

[holiman]

Looking into this lockup a bit, I was wondering why the sealBlock -> forkChoiceUpdated -> pool.Sync() was happening, and I found this comment

		// If the beacon chain is ran by a simulator, then transaction insertion,
		// block insertion and block production will happen without any timing
		// delay between them. This will cause flaky simulator executions due to
		// the transaction pool running its internal reset operation on a back-
		// ground thread. To avoid the racey behavior - in simulator mode - the
		// pool will be explicitly blocked on its reset before continuing to the
		// block production below.
		if simulatorMode {
			if err := api.eth.TxPool().Sync(); err != nil {
				log.Error("Failed to sync transaction pool", "err", err)
				return valid(nil), engine.InvalidPayloadAttributes.With(err)
			}
		}

Seems to me that adding the spinoff goroutines in this PR is basically a hack around this particular thing that is intentionally put there. I may be wrong, haven't fully grokked the entire situation yet.