Integer verbose (deprecate global options projpred.extra_verbose and projpred.verbose_project)

[fweber144]

This is a follow-up for PR #518: With #518 now being merged, this PR turns argument verbose of project(), varsel(), and cv_varsel() from logical to integer, which is common practice in R packages (see, e.g., argument silent of brms::brm()). This way, global options projpred.extra_verbose and projpred.verbose_project are now deprecated (which is desired, as they made it too complicated to achieve the desired verbosity).

File NEWS.md is unchanged here. I will update it as soon as #511 (or #508) has been merged. I would propose to add the following entry under "Major changes":

* Argument `verbose` of `project()`, `varsel()`, and `cv_varsel()` has been changed from logical to integer. However, logical values continue to work (since `as.integer()` is applied internally). Global options `projpred.extra_verbose` and `projpred.verbose_project` are now deprecated because additional verbosity can be achieved via higher integer values for argument `verbose`. The new global option `projpred.verbose` may be used to set argument `verbose` of `project()`, `varsel()`, and `cv_varsel()` globally. (GitHub: #519)

and to change the following entry:

* Minor enhancements concerning verbosity (e.g., the number of projected draws---resulting from clustering or thinning---is now printed out during the different steps of the computations). Also introduced global option `projpred.verbose` which may be used to set argument `verbose` of `varsel()` and `cv_varsel()` globally. (GitHub: #506)

(which was proposed to be changed to

* Several enhancements concerning verbosity (e.g., the number of projected draws---resulting from clustering or thinning---is now printed out during the different steps of the computations and verbose-mode output is redirected to `stderr()` instead of `stdout()`). Also introduced global option `projpred.verbose` which may be used to set argument `verbose` of `varsel()` and `cv_varsel()` globally. (GitHub: #506, #518)

in #518) to:

* Several enhancements concerning verbosity, e.g., the number of projected draws (resulting from clustering or thinning) is now printed out during the different steps of the computations and verbose-mode output is redirected to `stderr()` instead of `stdout()`. (GitHub: #506, #518)

[fweber144]

For illustration:

data("df_gaussian", package = "projpred")
df_gaussian <- df_gaussian[1:29, ]
dat <- data.frame(y = df_gaussian$y, df_gaussian$x)
rfit <- rstanarm::stan_glm(y ~ X1 + X2 + X3 + X4 + X5,
                           data = dat,
                           chains = 1,
                           iter = 500,
                           seed = 1140350788,
                           refresh = 0)

devtools::load_all(".")

### New global option `projpred.verbose` (which only serves the purpose of
### setting argument `verbose` of project(), varsel(), and cv_varsel() globally):
# options(projpred.verbose = 4)
# options(projpred.verbose = 0)

### Deprecated global option `projpred.extra_verbose`:
# options(projpred.extra_verbose = TRUE)
# options(projpred.extra_verbose = FALSE)
###

### Deprecated global option `projpred.verbose_project`:
# options(projpred.verbose_project = TRUE)
# options(projpred.verbose_project = FALSE)
###

prj <- project(rfit,
               predictor_terms = c("X1", "X2"),
               verbose = 2,
               seed = 46782345)

vs <- varsel(rfit,
             method = "forward",
             nclusters = 1,
             nclusters_pred = 1,
             refit_prj = TRUE,
             nterms_max = 1,
             verbose = 4,
             seed = 46782345)

prj_vs <- project(vs,
                  predictor_terms = c("X1", "X2"),
                  verbose = 2,
                  seed = 46782345)

prj_vs <- project(vs,
                  nterms = c(0, 1),
                  verbose = 2,
                  seed = 46782345)

cvvs <- cv_varsel(rfit,
                  nloo = 4,
                  method = "forward",
                  nclusters = 1,
                  nclusters_pred = 1,
                  refit_prj = TRUE,
                  nterms_max = 1,
                  verbose = 4,
                  seed = 46782345)

cvvs <- cv_varsel(rfit,
                  cv_method = "kfold",
                  K = 2,
                  method = "forward",
                  nclusters = 1,
                  nclusters_pred = 1,
                  refit_prj = TRUE,
                  nterms_max = 1,
                  verbose = 4,
                  seed = 46782345)