Add iceberg marker to test (#913)

BSE-5132: prepare_dataset (#873)

don't run conda search

Fix azure iceberg account name and storage account key (#862)

Pyarrow uses different names for storage account name and key than us but we should rely on their initialization function for robustness

Upper bound numba version (#835)

Split DF Tests on Azure CI (#787)

Improve LLM embed and generate APIs (#738)

Downgrade mpich to 4.1.2 for performance (#733)

Newer version of mpich introduced a performance regression in Spawn/Scatterv

NYC taxi changes (#658)

First round of changes for running NYC taxi and collecting performance data:

    Prevent unnecessary fallback when input DataFrame or parquet contains timestamp with non ns precision
    Optimize _init_lazy_distributed_arg by making sure it's only called once per LazyPlan.
    Adds cache flag to read_csv.


Co-authored-by: Todd A. Anderson <[email protected]>

Fix PyIceberg integration issues (#646)