BUG: Integrated pandas can't Read CSV while latest pandas can

[charliedream1]

Describe the bug

• Problem 1: a 25G csv file, latest pandas can load properly, however, "import xorbits.pandas as pd" can't, xorbits gives out EOF error
• Problem 2: a data frame data loaded from latest pandas can't be send to dup function (from xorbits.experimental import dedup)
• Problem 3: dedup function can handle a str with very long str, e.g. length between 4000-100,000, it gives out error "too many open files"

To Reproduce

To help us to reproduce this bug, please provide information below:

1. Your Python version: 3.10
2. The version of Xorbits you use: 0.6.3
3. Versions of crucial packages, such as numpy, scipy and pandas: numpy 1.26.0, scipy 1.11.3, pandas 2.1.1
4. Full stack of the error.
5. Minimized code to reproduce the error.

Expected behavior

A clear and concise description of what you expected to happen.

Additional context

Add any other context about the problem here.

[codingl2k1]

Problem 1: Is your csv file located in local disk or remote (by a url)?
Probelm 2: Are you using pandas to load the csv and constructing a xorbit Dataframe by the pandas Dataframe? If so, it could be out of memory crash, because the full data will be serilialized to worker.
Problem 3: The too many open files can be fixed by configure the ulimit.