Is the plot elegibility deterministic, could this enable huge cold storage plots power savings?

extending a bit on related uptime question:
Farmer uptime and chance to win - #4 by mxfh

There seems to be a serious flaw in the PoC green argument, why potentially rule out cold storage at all, with the same arbitrary 30 second cutoff for all plot sizes. Why not allow multiples of this lookup time for higher k values. Would take longer to determine a challenge winner, but I don’t see why this has to be immediate.

Wondering, if the protocol could be tweaked to allow this, to enable a significantly greener stable netspace phase chia:

Are the plot id groups per 1/512 chance stable?

If so, farmers could specialize on a fraction of the challenges. In the extreme, if you go all in on the 1/512 chance, you would only have to power up cold storage twice a week (10m * 512 equals about 3.5 days), with only minimal power consumption for the node looking for matching challenge ids.

You need timely challenges so that blocks continue to be created, so transactions occur. If we allowed cold storage… would that mean I may send a transaction that takes a week to finalize… because I need to give some farmer time to go grab a tape from his basement next Tuesday, and bring the correct bank of plots online?

Im thinking about a 20 seconds to a few minutes intervall in practice to spin up hdds / wake up relevant harvester nodes.
k32 keeps having 30 seconds
k33 has 1 minute
k34 2 and so on …
Reasonable cutoff would be 8 minutes then, so challenge is guaranteed to be determined within the 10 minute window.

The time limit is there to prevent just in time computation / storage tradeoffs. Expect this limit to become more and more stringent over time as both hardware and software advances, not more relaxed.

Then why not just have a shrinking base factor, that is multiplied by 2^k
Would make this transition gradual and keep the flexibility.