feat: lp indexing

[vacekj]

Describe your changes

implements lp positions and execution indexing

TODO from reviews:

• We want to store the trading pair likely as a new derived type (tuple) of (BYTEA, BYTEA)
• We definitely need a table storing the current state of any given position, including permanent information like the trading pair, price, fees.
• We need to include the reserve amounts in the lp_updates table, to have a history of the reserves over time.
• Price info, for both updates and current state

Issue ticket number and link

fixes #4739

Checklist before requesting a review

• 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:

indexer-only changes

[cronokirby]

This is a good start, but I think we want to compute some more derived information for use in the explorer, for example the price, so that the explorer can easily sort and fetch by price.

[cronokirby]

Let's clean up the accidental formatting changes and make sure that we have all the information we need to implement the APIs.

[conorsch]

This is a good start, but I think we want to compute some more derived information for use in the explorer

Downgrading to "draft", mostly to signal that we don't intend to block on upcoming point-release until this lands. Feel free to promote back to ready for review whenever, @vacekj!

[cronokirby]

A few more things I noticed:

• We want to store the trading pair likely as a new derived type (tuple) of (BYTEA, BYTEA)
• We definitely need a table storing the current state of any given position, including permanent information like the trading pair, price, fees.
• We need to include the reserve amounts in the lp_updates table, to have a history of the reserves over time.

[vacekj]

@cronokirby I addressed your feedback:

• Reserve amounts are already included in the table and events, including reserve amounts pre and post execution.
• I've added a new TradingPair type to postgres and made relevant changes
• The formatting in the sql file was intentional. If necessary, I'm open to reverting to the previous one.
• Fixed an oversight in the inserts where I was not passing in the actual data correctly
• added a table storing current state of positions, implemented as a view over the lp_updates table

Question about #: I might be misunderstanding, but the price here should just be reserves_1/reserves_2 right? I think we can calculate this on the fly in queries, but I can also add it to the view if necessary.

let me know how this looks to you @conorsch @cronokirby

also, who else should i be tagging ƒor reviews? don't wanna overload just aubrey, conor and lucas 😃

[cronokirby]

The problem with calculating the price as based on reserves is that it changes over time, as the reserves change, but in fact the price is static. It's based on https://buf.build/penumbra-zone/penumbra/docs/main:penumbra.core.component.dex.v1#penumbra.core.component.dex.v1.BareTradingFunction as present in the position.

If you're trading asset1 into asset2, the price is p / q, and if you're doing the opposite, it's q / p. For the effective price, which is what the dex explorer uses, you also need to take fees into account, so you get a worse price.

[vacekj]

@cronokirby this leads me to believe we then have all we need in the data? Can we proceed to review and merge?

[cronokirby]

I think we want:

• to store the price directly so that we can easily replicate the search by price queries
• to actually materialize each LP rather than computing it by taking the last seen update, because we need extra information like the price, and can avoid storing information that doesn't change in every update

but more broadly with this one we can actually test that it works by modifying the dex explorer raw indexing database queries to use it, and we should do that.

[cronokirby]

It would also be nice if we didn't create a diff over the parts of the database we haven't changed, we can format in a separate PR

[cronokirby]

Looks good to me now, did some brief testing as well