Support for fixest_multi

[etiennebacher]

fixest has a great interface to make multiple estimations in the same function call. It returns an object of class fixest_multi which is basically a list of fixest objects.

This PR adds support for fixest_multi for the following functions: check_outliers(), check_overdispersion(), r2(), model_performance().

Important: there were some discussion in #429 about which checks should be implemented for this package. I only looked at the functions supporting fixest and added support for fixest_multi, that's it (+ a few tests)

Examples:

library(performance)
library(fixest)
#> 
#> Attachement du package : 'fixest'
#> L'objet suivant est masqué depuis 'package:performance':
#> 
#>     r2

base <- iris
names(base)  <- c("y1", "y2", "x1", "x2", "species")
res <- feols(c(y1, y2) ~ x1 + csw(x2, x2^2) | species, base)
model_performance(res)
#> $`lhs: y1; rhs: x1 + x2`
#> # Indices of model performance
#> 
#> AIC     |     BIC |    R2 | R2 (adj.) | R2_within | R2_within_adjusted |  RMSE | Sigma
#> --------------------------------------------------------------------------------------
#> 106.231 | 121.284 | 0.837 |     0.832 |     0.572 |              0.566 | 0.333 | 0.339
#> 
#> $`lhs: y1; rhs: x1 + x2 + I(x2^2)`
#> # Indices of model performance
#> 
#> AIC     |     BIC |    R2 | R2 (adj.) | R2_within | R2_within_adjusted |  RMSE | Sigma
#> --------------------------------------------------------------------------------------
#> 107.743 | 125.807 | 0.837 |     0.832 |     0.573 |              0.564 | 0.333 | 0.340
#> 
#> $`lhs: y2; rhs: x1 + x2`
#> # Indices of model performance
#> 
#> AIC    |    BIC |    R2 | R2 (adj.) | R2_within | R2_within_adjusted |  RMSE | Sigma
#> ------------------------------------------------------------------------------------
#> 65.423 | 80.476 | 0.551 |     0.539 |     0.251 |              0.240 | 0.291 | 0.296
#> 
#> $`lhs: y2; rhs: x1 + x2 + I(x2^2)`
#> # Indices of model performance
#> 
#> AIC    |    BIC |    R2 | R2 (adj.) | R2_within | R2_within_adjusted |  RMSE | Sigma
#> ------------------------------------------------------------------------------------
#> 66.637 | 84.701 | 0.553 |     0.538 |     0.255 |              0.239 | 0.290 | 0.296

[etiennebacher]

Failures seem unrelated (happen in test-cronbachs_alpha)

[strengejacke]

There will be a lot of failures in tests, because some functions rely on changes in upstream packages (inisght, parameters), which will resolve after this release cycle.

[strengejacke]

Thanks! We may think of adding an additional class for specific print-methods, but the printed lists already look quite ok for now.