Make validTimePeriod configurable

[ali-behjati]

I will not deploy this change to testnet until we know the proper value, and that mean we can measure time from a pyth published to a client subscribed to the price service getting that price.

The ideal value is derived like this:

2*price_update_frequency + e2e(price published, a price service client receives it) + 2*target_network_latency + target_network_block_time

In solana we have a ~12 seconds validity time threshold which was set for the congestion periods. We might also consider setting in acceptable minimum here, I can change the contract to instead of being initialized with 0, be initialized with that value.

[jayantk]

change is generally fine. I think the configured validity times are likely too low though (see inline)

[jayantk]

Can we set the initial time periods to err on the high side? I think some of these thresholds are likely too aggressive -- would suggest doubling all of them as a baseline, and using 60 seconds as a minimum. (Note that your calculation also doesn't account for clock drift between the chains, which could be substantial)

I suggest this because we are currently not in a position to measure how many failures occur because the threshold is too aggressive. Without that measurement, we could end up in a situation where Pyth is extremely flaky on a target chain, and we wouldn't know it. We can optimize these thresholds later once our measurement infrastructure is in place.

[ali-behjati]

These are my initial estimates based on the block time of the given chain. But we can measure it once we deploy attester in Pythnet.

I shared my opinion about clock drift in the above thread too. I don't think it be an issue.

[ali-behjati]

I applied your suggestion.

[jayantk]

Also, no need to overcomplicate this by putting in a minimum validity threshold. We can guard against that in code that's easier for us to change, especially since we don't know the right minimum.

[jayantk]

not sure about this trading status change, but LGTM aside from that