ingest different df types + tests

[artiom-matvei]

No description provided.

[artiom-matvei]

@vincentarelbundock This one should do the trick for issue #108
These tests were run manually to make sure it accepts pandas and duckdb

import polars as pl
import statsmodels.formula.api as smf
from marginaleffects import *
import numpy as np
import pandas as pd
penguins = pd.read_csv(
    "tests/data/penguins.csv",
    na_values="NA"
).dropna()
mod = smf.ols(
  "body_mass_g ~ flipper_length_mm * species * bill_length_mm + island",
  penguins,
).fit()
p = predictions(mod, newdata=penguins)

import polars as pl
import statsmodels.formula.api as smf
from marginaleffects import *
import numpy as np
import pandas as pd
penguins = pd.read_csv(
    "tests/data/penguins.csv",
    na_values="NA"
).dropna()
import duckdb
df_duckdb = duckdb.sql("SELECT * FROM penguins LIMIT 10")
mod = smf.ols(
  "body_mass_g ~ flipper_length_mm * species * bill_length_mm + island",
  penguins,
).fit()
p = predictions(mod, newdata=df_duckdb)

[vincentarelbundock]

pandas is still used in hypotheses_joint.py. Once that is replaced, pandas should be placed in the "extras" list of pyproject.toml, and ideally pyarrow as well, unless we have a compelling reason to keep it there.

[artiom-matvei]

it seems like obj is of type model and it shouldn't ever be of type dataframe. It was introduced here #94 but not sure why. I also checked in R and obj is not a dataframe in R either.

If it were to be a dataframe all the following operations would fail so I think we can remove it. @vincentarelbundock do you agree?

[vincentarelbundock]

That seems right but I won't have time to dig deep, so I'll trust you here.

[artiom-matvei]

@vincentarelbundock please review and don't hesitate to share your comments.
When merged, please, let me know what would be the next steps. Thank you!