Create ScalarUDF Benchmark

[crepererum]

What

Create a ScalarUDF benchmark.

Why

If we want to improve performance (see #27), we should know if things work or not.

How

Create the two UDFs:

• add_one (integer test): takes i64 and adds 1, returns it
• sub_str (complex type test): takes a string and splits it at the first ., returns the 2nd half

And implement it in 3 different ways:

• native UDF: Just implement it was a "normal" UDF w/o any WASM. This will be the baseline.
• Rust Guest: Use the native version but within WASM as a Rust guest. This helps us to estimate the WASM overhead (both in terms of JIT performance and the IO overhead).
• Python Guest: Implement a simple Python function. This is as bad as it gets, but we need to know how slow that is.