Add E2E tests for acceleration engines types support

[sgrebnov]

Motivation/Why?

• We have a record of issues related to acceleration backends when specific data type can't be saved locally (runtime panic/error) or query failure.
• This PR attempts to test all possible types support by acceleration backends by creating local DuckDB instance as a data source containing different types (DuckDB was selected as the one easy to setup locally; if we like this approach we can switch to remote s3/parquet file), enable supported accelerations and compare query results (query source DuckDB) vs accelerated backend (DuckDB, Arrow, sqlite, postgreSQL(not currently covered))

Implements Github workflow to test that acceleration backend correctly support common types.

1. Creates test duckdb database and populates it with main Arrow types (isolated table per type as some types cause Spice runtime to crash so all types are isolated in own table)
2. For each type creates spicepod and runs query twice

• without acceleration
• with local acceleration

3. Compares results and creates summary table

Example test run (scroll page down to find summary): https://github.com/spiceai/spiceai/actions/runs/8872467039

Sample report for boolean