compilation fails with pandas >=2.0

[a4rcvv]

Describe the bug

Sometimes scmlv fails to compile when I use it with pandas>=2.0.

To Reproduce
Steps to reproduce the behavior:

1. Prepare a log folder containing negotiations.csv with None in the agreement column (i.e. logs containing broken negotiations).
2. Run scmlv show -f <log folder>
3. Parse Error occurs, and the visualizer raises TypeError

# Error of compilation
No tournament found ... reading world info
        Parse Error logs
        Did not find agent_stats
        Did not find product_stats
        Did not find world_stats
        Did not find contracts
        Did not find contract_stats
        Did not find negotiations
        Did not find offers
        Did not find neg_stats
        Did not find breaches

# Error from the visualizer
Traceback (most recent call last):
  File "/Users/a4rcvv/Documents/projects/scml-repro/.venv/lib/python3.11/site-packages/streamlit/runtime/scriptrunner/script_runner.py", line 565, in _run_script
    exec(code, module.__dict__)
  File "/Users/a4rcvv/Documents/projects/scml-repro/.venv/lib/python3.11/site-packages/scml_vis/presenter.py", line 1548, in <module>
    main(folder)
  File "/Users/a4rcvv/Documents/projects/scml-repro/.venv/lib/python3.11/site-packages/scml_vis/presenter.py", line 193, in main
    products["product"].unique(),
    ~~~~~~~~^^^^^^^^^^^
TypeError: 'NoneType' object is not subscriptable

Expected behavior

scmlv can compile log files successfully.

System (please complete the following information):

• scml-vis version: 0.3.1
• Python version: 3.11
• OS: macOS 14.2

Additional context

One of the current workaround is to use pandas<2.0.0.

Pandas treats a string of "None" as NaN in default since 2.0.0(ref: pandas-dev/pandas#50286, https://pandas.pydata.org/docs/dev/whatsnew/v2.0.0.html).

But it seems that scmlv wants to treat "None" as "None" as a string at here. When I use pandas>=2.0, NaN may be provided to this function which causes an error.