v2: remove deprecated legacy tables superseded by source/behavior_fixcost

v2: rename source parquet files to user-friendly names under v2/source/

v2 trading release: add v2/source/source_metadata.json

v2 trading release: add v2/benchmark_directgen/benchmark_metadata.json

v2 trading release: add v2/benchmark_directgen/test.jsonl.gz

v2 trading release: add v2/benchmark_directgen/validation.jsonl.gz

v2 trading release: add v2/benchmark_directgen/train.jsonl.gz

v2 trading release: add v2/splits/wallet_ids_experiment_2000.txt

v2 trading release: add v2/splits/split_metadata.json

v2 trading release: add v2/splits/test.parquet

v2 trading release: add v2/splits/validation.parquet

v2 trading release: add v2/splits/train.parquet

Upload V2 merged table with fixed cost basis

docs(v2): add explicit 2-hour kline window semantics for k0-k5

Add v1 data description for raw/direct/profilegen/retrievalgen support set

Add clearer raw data filename: raw/data_v1.parquet

Add retrievalgen support set v1 (exclude 2000 experiment wallets)

Add v1 profilegen LLM dataset (train/validation/test + splits + state)

Add v1 direct benchmark split files (train/validation/test + split metadata)

Add files using upload-large-folder tool

Upload V2 behavior base table (medium)

Upload V2 market semantic table (small)

Add V2 release README

Remove old non-time parquet to avoid duplicate row counting

Add time-enriched mk10plus parquet (non-overwrite upload)

Upload V1 cognition mk10plus minimal parquet dataset

initial commit