Initial testing of formulaic's i operator

[leostimpfle]

Have started testing formulaic's implementation of the i operator (matthewwardrop/formulaic#265) related to #782

Once the formulaic side is ready, this should largely be a drop-in replacement of our current implementation with some tweaks to the syntax, see for example here:

pyfixest/tests/test_i.py

Lines 199 to 221 in dcf3d47

	@pytest.mark.parametrize(
	"fml_py, fml_r",
	[
	("dep_var ~ i(C(state, reduce_rank=True))", "dep_var ~ i(state)"),
	(
	"dep_var ~ i(C(state, contr.treatment(1), reduce_rank=True))",
	"dep_var ~ i(state, ref = 1)",
	),
	("dep_var ~ i(C(state, reduce_rank=False), year)", "dep_var ~ i(state, year)"),
	(
	"dep_var ~ i(C(state, contr.treatment(1), reduce_rank=True), year)",
	"dep_var ~ i(state, year, ref = 1)",
	),
	(
	"dep_var ~ i(C(state, reduce_rank=False), year) | state",
	"dep_var ~ i(state, year) | state",
	),
	(
	"dep_var ~ i(C(state, contr.treatment(1), reduce_rank=True), year) | state",
	"dep_var ~ i(state, year, ref = 1) | state",
	),
	],
	)

FYI @s3alfisc