feat(hogql): LRU cache for parsing

[mariusandra]

Problem

The HogQL lifecycle query takes ~3s to parse. About 2.5s of those 3s are spent waiting for antlr.
This is work that could only be done once and then cached.

There are other optimisations we could take in later stages, but let's tackle the biggest one first.

Changes

• This is more of an experiment and less of a PR to merge.
• Adds a simple LRU cache for the Antlr query parsing part.
• Note: just the first parsing part is cached. The following parts (resolver, printer, etc) are not.
• This will cache the query per web pod thread. With several dozen of those in operation (not sure of the exact number), we're still doing too much work. A redis-based cache might be better.
• If we use placeholders and compose our queries in the right way, this can yield a big boost. Each query part can be cached independently.

Alternatives.

• Find a way to make the antlr parser a lot faster (this might be a quick win, I just don't know)
• Replace the antlr parser with a homegrown recursive descent parser (... in Rust for even more oomph)

How did you test this code?

Saw the lifecycle query load almost instantly on every second refresh.

[posthog-bot]

This PR hasn't seen activity in a week! Should it be merged, closed, or further worked on? If you want to keep it open, post a comment or remove the stale label – otherwise this will be closed in another week.