[python-package] [R-package] include more params in model text representation (fixes #6010)

[jameslamb]

Fixes #6010.

Proposes adding the following parameters to the set written to model text files (and therefore also the set preserved by pickling/unpickled in Python and saveRDS() / readRDS() in R):

data_sample_strategy
num_grad_quant_bins
quant_train_renew_leaf
seed
stochastic_rounding
use_quantized_grad

It also refactors the include/config.h mechanism, to make the relationship between that file, src/config_auto.cpp, and the Parameter docs clearer.

Notes for Reviewers

This PR was mainly inspired by #6017 (comment), but it's informed by many prior conversations related to this topic:

• Refine config object #1381
• [docs] generate parameters description from config file. Stage 1 #1409
• [docs] generate parameters description from config file. Stage 2 #1413
• incorrect logged parameters in model file #2208
• [python] save all param values into model file #2589
  • especially [python] save all param values into model file #2589 (comment)
• [python] [R-package] refine the parameters for Dataset #2594
• Load back saved parameters with save_model to Booster object #2613
• [R-package] centralize Dataset parameter aliases and add tests on parameter updating #2767
• [Docs] 4.0.0 parameters missing from model.txt #6010
• Fix updates in random forest model using GOSS data sample strategy #6017
  • especially Fix updates in random forest model using GOSS data sample strategy #6017 (comment)

[jmoralez]

LGTM. Just a suggestion

[jameslamb]

I tested both sides of this, like this:

.expect_in <- function(x, y) {
  if (exists("expect_in")) {
    expect_in(x, y)
  } else {
    missing_items <- x[!(x %in% y)]
    if (length(missing_items) != 0L) {
      error_msg <- paste0("Some expected items not found: ", toString(missing_items))
      stop(error_msg)
    }
  }
}

# testthat not loaded
.expect_in(c("a", "b"), c("d", "a", "a"))

Error in .expect_in(c("a", "b"), c("d", "a", "a")) :
Some expected items not found: b

# testthat loaded
library(testthat)
.expect_in(c("a", "b"), c("d", "a", "a"))

Error: x (actual) isn't fully contained within y (expected).

• Missing from expected: "b"
• Present in expected: "d", "a", "a"

[github-actions]

This pull request has been automatically locked since there has not been any recent activity since it was closed. To start a new related discussion, open a new issue at https:/microsoft/LightGBM/issues including a reference to this.