Peidian li

[MINOR] Shrink the flink-connector-runtime jar

6

### What changes were proposed in this pull request? - shrink the flink-connector-runtime jar. - Change client-java scope from implementation to CompileOnly in flink module. - use the client-java-runtime in...

- `distinct()` will storage the data, and the downstream will read from the shuffle. We do not need the cache any more. - Use the `count()` to instead the `collect()`...

**Is your feature request related to a problem? Please describe.** Is there a plan to support to read the apache iceberg?

range start index 2896120 out of range for slice of length 2896119

3

### Describe the bug ``` ell In[1], line 6 2 daft.context.set_runner_ray(noop_if_initialized=True) 4 df = (daft.read_json('/mnt/xx/*.jsonl') 5 .select(daft.col('metadata')['url']).into_partitions(5000)) ----> 6 df.write_parquet('/mnt//url/zh_url/') File /home/work/mount/python/user_env/lib/python3.10/site-packages/daft/api_annotations.py:26, in DataframePublicAPI.._wrap(*args, **kwargs) 24 type_check_function(func, *args, **kwargs) 25...