Adapt the nuisance estimation for the IV type score in the PLR model

[MalteKurz]

Description

In this PR the nuisance estimation for the IV-type score in the PLR model is adapted to be in line with the DML paper Chernozhukov et al. (2018).

• Results for the default score='partialling out' (Equation (4.4) in Chernozhukov et al. (2018)) are not affected by the changes in this PR. However, the naming of the predictions is changed from g_hat to l_hat to be better in line with Chernozhukov et al. (2018). Still the API is not changed to be backward-compatible, i.e., when initializing an DoubleMLPLR object, learners ml_g and ml_m need to be specified. The estimator ml_g is then used to estimate the effect of X on Y (i.e. l_0(X) = E[Y|X]) and the estimator ml_m to estimate the effect of X on D (i.e. m_0(X) = E[D|X]). When setting hyperparamters learner='ml_l' and learner='ml_m' should be used.
• For the score='IV-type' (Equation (4.3) in Chernozhukov et al. (2018)) the implementation now follows the approach described on pp. C31-C33 in Chernozhukov et al. (2018). This means that an initial estimate for theta_0 is obtained via the 'partialling out' score. Then an estimate for g_0(X) is obtained by regressing Y - theta_0 * D on X. For both nuisance functions l_0(X) (needed for the preliminary theta_0 estimate) and g_0(X), the same ML estimator (specified in ml_g) is used. However, hyperparamters can be set individually via learner='ml_l' and learner='ml_m'.

PR Checklist

• The title of the pull request summarizes the changes made.
• The PR contains a detailed description of all changes and additions.
• References to related issues or PRs are added.
• The code passes all (unit) tests.
• Enhancements or new feature are equipped with unit tests.
• The changes adhere to the PEP8 standards.

[MalteKurz]

I changed this PR back into a draft PR, as we first need to discuss whether we want to additionally adapt the API or not.