Fix slot checks in light client p2p spec

[michaelsproul]

I think the p2p spec for light client updates should require the received update to be newer than previously forwarded updates. In other words, the condition should be corrected to:

No other finality_update with a lower greater or equal finalized_header.slot was already forwarded on the network

However I think it's somewhat clearer to state the condition without a negation, as in:

[IGNORE] The finalized_header.slot is greater than the slot of all previously forwarded finality updates

[ralexstokes]

makes sense, otherwise there is a mild DoS vector where I just keep spamming the current update to add load to the gossip net

[dapplion]

Looks reasonable to me, CC: @etan-status

[etan-status]

Thanks, yes, that was the intention (slots must strictly monotonically increase in gossip), and is also how it is implemented in Nimbus. The new wording is clearer as well.