epic: Measure performance

[friendlymatthew]

It would be nice to create a set of dedicated scripts and benchmarks in hopes of improving certain UDF flows like variant_get or json_to_variant

This way, people interested can reuse this setup and focus on investigating perf wins

@XiangpengHao linked Clickhouse's JSON bench project which provides a nice realistic dataset

I plan on doing the following:

• Add a script to generate the dataset
• Write profiling scripts
• Set up benchmarks

[XiangpengHao]

I'm not sure if this will be helpful: https://github.com/XiangpengHao/liquid-cache-bench/tree/main/json_bench

(llm generated anyway😆)

[XiangpengHao]

Now I have something very concrete to look at, will try to do something later today.