pindexer: add dex block summary

[erwanor]

Describe your changes

WIP pairing

Issue ticket number and link

penumbra-zone/dex-explorer#338

Checklist before requesting a review

• I have added guiding text to explain how a reviewer should test these changes.

• If this code contains consensus-breaking changes, I have added the "consensus-breaking" label. Otherwise, I declare my belief that there are not consensus-breaking changes, for the following reason:

  REPLACE THIS TEXT WITH RATIONALE (CAN BE BRIEF)

[conorsch]

@JasonMHasperhoven and I paired a bit to debug the failing smoke test. To investigate, I pulled down this branch, ran just dev to spin up the fullnode indexing config, and quickly identified the specific error that pindexer was showing:

That error, in text, is:

Error: error returned from database: column "batch_swaps" is of type batch_swap_summary[] but expression is of type jsonb
Caused by:
    column "batch_swaps" is of type batch_swap_summary[] but expression is of type jsonb

Clear enough: we've got to massage the types a bit. Typically we use JSON blobs to store Penumbra types in postgres via pindexer, but as the spec in penumbra-zone/dex-explorer#338 indicates, we don't want to do that here. Instead, we want a full custom type inside the db. I took a stab at implementing sqlx::Encode for the BatchSwapSummary type, but wasn't able to finish it. The build-time error I'm getting now is:

error[E0277]: the trait bound `for<'a> &'a [BatchSwapSummary]: sqlx::Encode<'_, Postgres>` is not satisfied
    --> crates/bin/pindexer/src/dex_ex/mod.rs:1270:16
     |
1270 |         .bind(&batch_swap_summaries)
     |          ----  ^^^^^^^^^^^^^^^^^^^^ the trait `for<'a> sqlx::Encode<'_, Postgres>` is not implemented for `&'a [BatchSwapSummary]`, which is required by `&Vec<BatchSwapSummary>: sqlx::Encode<'_, Postgres>`

Subsequent work should pick up from there. Take the Encode impl with a grain of salt, @JasonMHasperhoven , because this code isn't actually working yet! But hopefully this is one step closer.

[erwanor]

@cronokirby noticed that some inserts fail if we are adding a UNIQUE constraint on height, but my understanding is that this part of the pipeline should have exactly once semantics on delivery. is that right?

wondering if this is caused by the catchup logic: https://github.com/penumbra-zone/penumbra/blob/erwan/pindexer_block_summary/crates/util/cometindex/src/indexer/indexing_state.rs#L271 (<= vs. <). unsure.

[erwanor]

AFAICT this is good to go, the integration test seems to timeout because of the commitment source addition.