[query] from_pandas needs to support Pandas missing types and support numpy nums better

[johnc1231]

Pandas missingness is crazy. pd.isna tells you if something is "missing", which means either NaN, or a special NA sentinel value. For floats they use NaN, and specifically for the pandas special Int64DType and Int32DType and nothing else they use this NA value.

They don't have an easy way to distinguish between NA and NaN, so I first check if something isna, then check if it's a float to differentiate between the cases.

They also don't have a way to test if something is a Int32DType for some reason. So I use is_int64_dtype, and if that fails I fall back to is_integer_dtype, which is true for both Int64DType and Int32DType.