Switch to pandas 2.0.0rc1

Seems to work fine with the new PyArrow datatypes.

csv_metadata_quality/app.py

@@ -73,8 +73,7 @@ def run(argv):
     # set the signal handler for SIGINT (^C)
     signal.signal(signal.SIGINT, signal_handler)
 
-    # Read all fields as strings so dates don't get converted from 1998 to 1998.0
-    df = pd.read_csv(args.input_file, dtype=str)
+    df = pd.read_csv(args.input_file, dtype_backend="pyarrow")
 
     # Check if the user requested to skip any fields
     if args.exclude_fields:

pyproject.toml

@@ -12,7 +12,7 @@ csv-metadata-quality = 'csv_metadata_quality.__main__:main'
 
 [tool.poetry.dependencies]
 python = "^3.9"
-pandas = "^1.5.2"
+pandas = "2.0.0rc1"
 python-stdnum = "^1.18"
 requests = "^2.28.2"
 requests-cache = "^0.9.8"