fix: Hash ancillaryData in entity IDs to prevent PostgreSQL index overflow

[md0x]

Fixes UMA-3008

Subgraph fails at block 81660211 with error: index row size 3096 exceeds btree version 4 maximum 2704. Long ancillaryData from Polymarket resolution criteria causes OptimisticPriceRequest entity IDs to exceed PostgreSQL's btree index limit.

Solution: Changed the ID construction to hash the ancillaryData using keccak256, producing a fixed-length 66-character hash.

[linear]

UMA-3008 Hash ancillaryData in entity IDs to prevent PostgreSQL index overflow

[gsteenkamp89]

This looks good.
My only concern is that if any of these get disputed, they will move into the voting contract, so the voting subgraph will suffer from the same problem.

We should make similar changes to the voting subgraph's ID generation as well.
While we're at it, we can do the same for the OOv2 subgraph.

[gsteenkamp89]

I have verified that

1. Nowhere in the oracle dapp do we attempt to query a single entity by constructing it's ID ✅
2. The Voting dapp does do this, so we would have to ensure we mimic the logic here to construct the ID. ❌

[gsteenkamp89]

Looks good!
Thanks for getting this in.