Asset-Hub Migration: Investigate throughput

[ggwpez]

Estimate how many messages and how much time we will need to migrate all pallets to the AH.

First investigations were promising with the indices pallet. Now going to test with real state on AH to measure the PoV sizes in the collator logs. My approach currently is to migrate pallets and then see how fast it is. Pallet-indices was very fast, next going to try pallet-balances.

Issues identified so far:

•  DMP delivery fee increases once the relay starts spamming
• AH can silently drop messages if they are too big. Need to safeguard against this on the relay side already.

[burdges]

You could go the other way: Fork the relay chain into a parachain, upgrade that fork into AssetHub v2, with a migration, and the message over all the AssetHub v1 accounts.

I'd honestly just not think of this as messaging though: Freeze realy chain accounts, copy that state root to AssetHub as a subtree, do a state migration in AssetHub from that subtree into AssetHub accounts, and O(1) drop that state tree when done. All told only one "message": the state root hash.