Move from dealbot subgraph to ponder

## Done Criteria
Instead of relying on https://github.com/FilOzone/dealbot/tree/main/apps/subgraph, we use ponder for finding anonymous pieces (meeting the needs outlined in https://github.com/FilOzone/dealbot/blob/main/apps/subgraph/README.md

## Why Important
Less operations potentially of dealbot needing to rely Goldsky
Less cost since most this data is already stored/indexed in FilOzone/infra anyway for early repair.

## Notes
### 202606 Notes from Kuba
https://filecoinproject.slack.com/archives/C08TVNKJV7C/p1782246908163489
* If we want to rely on Ponder beyond analytics, we should definitely go with an HA PostgreSQL cluster.
- Note: this is follow on work to https://github.com/FilOzone/infra/issues/253
* My strong suggestion is to do that only in Prod and deploy both networks there, rather than using staging for calibration.
- https://github.com/FilOzone/dealbot/issues/297
* How would Dealbot use it: It would create its own ponder indexing rules, creating tables with columns that it needs. If you need stored procedures/views that can be done on top (managed outside of Ponder).
* We can probably share bits and pieces across foc-observer/early-repair/dealbot ponders, but each project wants a slightly different view of the underlying data.  So my assumption is that sharing would be limited to library code for decoding, rather than sharing tables

Start with dealbot's apps/subgraph/README.md to understand the anonymous-piece requirements, then review the related deployment and infrastructure context in infra#253 and dealbot#297. Identify how Ponder indexing rules would provide the required view, including the PostgreSQL and production deployment constraints. Done means dealbot no longer relies on the dealbot subgraph for finding anonymous pieces.

Written by the indexing model from the issue text.