[Discussion]: Support multiple choices/select specs based on dataset

### Feature description

Goal: allow to support different variables/values choices/select based on the dataset being used

This is a feature that existed on `teal.transform`, but hasn't been matched by `teal.picks`

The previous API used a list of `data_extract_spec** to provide with each dataset specification and that may not be the best approach for picks.

### Proposal do discuss

- Picks specification as it is right now provides with defaults/fallback `variables`/`values`
- Dataset specific:
1. Named dataset arguments provide with dataset-specific `variables`/`values`
2. Post-picks operations to provide dataset-specific

Mock example exploring all possible options:
- Has 3 datasets
- `ADTTE` does not have a custom specification and only uses `variables()`
- `ADSL` selects `SEX` variables and has `values()`
- `ADRS` selects `AGE` variable and does not have `values()`

My concern with first approach below is that we would have repetitive parameters names.

It's is possible, but a concern nonetheless (click here to expand)

```r
foo <- function(...) {
dots <- rlang::list2(...)

datasets = Filter(function(x) inherits(x, "datasets"), dots)
variables = Filter(function(x) inherits(x, "variables"), dots)
values = Filter(function(x) inherits(x, "values"), dots)
print("Datasets")
print(datasets)
print("Variables")
print(variables)
print("Values")
print(values)
}

foo(datasets(), variables(), ADSL = variables(), ADSL = values())
#> [1] "Datasets"
#> [[1]]
#>
#> choices: tidyselect::everything()
#> selected: 1L
#> multiple=FALSE, ordered=FALSE, fixed=FALSE
#>
#> [1] "Variables"
#> [[1]]
#>
#> choices: tidyselect::everything()
#> selected: 1L
#> multiple=FALSE, ordered=FALSE, fixed=FALSE, allow-clear=FALSE
#>
#> $ADSL
#>
#> choices: tidyselect::everything()
#> selected: 1L
#> multiple=FALSE, ordered=FALSE, fixed=FALSE, allow-clear=FALSE
#>
#> [1] "Values"
#> $ADSL
#>
#> choices:
#> selected:
#> multiple=TRUE, ordered=FALSE, fixed=FALSE
```

```r
picks(
datasets(c("ADSL", "ADRS", "ADTTE")),
variables(),
ADSL = variables(selected = c(SEX)),
ADRS = variables(selected = c(AGE)),
ADSL = values(selected = "F")
```

This seems much more R-like _(functions names TBD)_

```r
picks(
datasets(c("ADSL", "ADRS")),
variables()
) |> picks_extra_variables(
ADSL = variables(selected = c(SEX)),
ADRS = variables(selected = c(AGE))
) |> picks_extra_values(
ADSL = values(selected = "F")
)
```

### Previous API

```r
variables = list(
teal.transform::data_extract_spec(
dataname = "ADSL",
select = teal.transform::select_spec(
label = "Select variables:",
choices = teal.transform::variable_choices(data[["ADSL"]]),
selected = c("AGE", "RACE", "SEX"),
multiple = TRUE,
ordered = TRUE,
fixed = FALSE
)
),
teal.transform::data_extract_spec(
dataname = "ADRS",
filter = teal.transform::filter_spec(
label = "Select endpoints:",
vars = c("PARAMCD", "AVISIT"),
choices = teal.transform::value_choices(data[["ADRS"]], c("PARAMCD", "AVISIT"), c("PARAM", "AVISIT")),
selected = "INVET - END OF INDUCTION",
multiple = TRUE
),
select = teal.transform::select_spec(
label = "Select variables:",
choices = teal.transform::variable_choices(data[["ADRS"]]),
selected = c("AGE", "AVAL", "ADY"),
multiple = TRUE,
ordered = TRUE,
fixed = FALSE
)
)
)
```

### Code of Conduct

- [x] I agree to follow this project's Code of Conduct.

### Contribution Guidelines

- [x] I agree to follow this project's Contribution Guidelines.

### Security Policy

- [x] I agree to follow this project's Security Policy.

The issue names no files or tests. Start by reviewing the current teal.picks picks specification and the previous teal.transform data_extract_spec API, then compare the two proposed dataset-specific designs; done means an agreed implementation direction for dataset-specific variables and values choices.

Written by the indexing model from the issue text.