Podlozhnyy-module

1. Update `regression` package in the part of `holoviews` package for smooth running with latest Python versions
2. Minor fixes in `regression` package

**Full Changelog**: https://github.com/NPodlozhniy/podlozhnyy-module/compare/2.5.1...2.5.2

Within the release more statistics functionality is added!

1. 🆎 `abtesting` for conversion (Ztest) and average (Ttest) targeted experiments, including its multivariate versions and Dunnett's correction for both scenarios. An undeniable advantage of the implementation is that it expects means and standard deviations as an input, hence you can leverage various techniques before applying the test, such as delta-method for ratio-metrics, linearization, reweighting, etc - generally any sophisticated kind of variance calculation and then feed it into the criterion
2. `anova` (one-way only) method that is the classical choice in case of multiple groups for one dependent variable and one independent factor, firstly it has the same plus as `abtesting` module - it's enough to supply just the lists of means and deviations, no need for the source samples and what is more it includes pairwise Tukey HSD test - the most accurate and widely used procedure for post-hoc analysis
3. `regression` aggregated tables processing in the same version as it was described in [2.3-alpha](https://github.com/NPodlozhniy/podlozhnyy-module/releases/tag/2.3-alpha) version, dynamic functions adjustment is still in to do list


**Full Changelog**: https://github.com/NPodlozhniy/podlozhnyy-module/compare/2.2.1...2.5.1

🅰️NOVA module is created!
In addition to a classical ANOVA it implements pairwise Tukey HSD test for post-hoc analysis.
Demo is available in the [playbook](https://github.com/NPodlozhniy/podlozhnyy-module/blob/main/Playbook.ipynb)


**Full Changelog**: https://github.com/NPodlozhniy/podlozhnyy-module/compare/2.4-beta...2.5rc0

🆕 AB testing package for conversion-like metrics is added!
In addition to a classical Z-test it includes Dunnett's correction.
Usage details are available in the [playbook](https://github.com/NPodlozhniy/podlozhnyy-module/blob/main/Playbook.ipynb)
For more details on Dunnett's test check out the dedicated article [Dunnett's Correction for ABC testing](https://npodlozhniy.github.io/posts/dunnett-correction/)

**Full Changelog**: https://github.com/NPodlozhniy/podlozhnyy-module/compare/2.3-alpha...2.4-beta

Pre-release functionality

:fire: Limited **aggregated tables processing** for regressions risk analysis, rolled-out for static functions only as of now, namely `woe`, `IV`, etc. For dynamic functions the upgrades are coming with the next stable release. It helps to perform the analysis in case of the data has been grouped and supplied with the `grouping_size` column representing the number of observations per row of aggregated table, while `target` variable will mean not a _bool_ flag but the total _integer_ number of conversions per group

:ballot_box_with_check: Formatting fix for bucket's naming, now the small numbers are displayed in a more user-friendly way


**Full Changelog**: https://github.com/NPodlozhniy/podlozhnyy-module/compare/2.2.1...2.3-alpha

Correlations package update

1. The functionality which was relevant before only to numeric features, such as `correlation_significance` is extended for categorical variables
2. Now `plot_corr_matrix` is more a general interface to visualize the output correlation matrix, including apart from coefficient itself, their statistical significance as well

**Full Changelog**: https://github.com/NPodlozhniy/podlozhnyy-module/compare/2.1...2.2.1