Transparently expose different methods for finding dependencies

[nuest]

The user should be able to select which methods are applied to find the dependencies which are put in the Dockerfile. These might be combined or follow each other, and even depend on each other

• parsing-based (with automagic)
• execution-based (for scripts and R Markdown files; might need previous
• sysreqs-based (for finding system dependencies)
• Add option to use hadley/requirements instead of document execution #176

These different steps should have a common API and be chainable, i.e. be able to rely on the previous results.

df <- dockerfile(from = "sessionInfo.Rdata", methods = c(cntnrt_parse, cntnrt_execute, cntnrt_sysdeps), ...)

Internally this might result in

df <- cntnrt_parse(from)
df <- cntnrt_execute(df, from, preinstallAlreadyDetectedPackages = TRUE)
df <- cntnrt_parse(df, from, goOnline = TRUE)

See also ideas in #101

---

Can also have one option cntnrt_all that creates a union of all options, and cntnrit_parse should be a union of cntnrt_automagic and cntnrt_requirements:

> library("requirements")
> require
require           requireMethods    requireNamespace  requirements::    
> requirements::req_
requirements::req_file       requirements::req_code       requirements::req_namespace  
> requirements::req_file("main.Rmd")
 [1] "rmarkdown"       "sdmpredictors"   "randomForest"    "summarytools"   
 [5] "Boruta"          "caret"           "raster"          "ggplot2"        
 [9] "GGally"          "rgdal"           "corrplot"        "splitstackshape"
[13] "e1071"          
> library("automagic")
> automagic::
automagic::automagic               automagic::install_package_guess   automagic::get_package_details     automagic::install_deps_file       
automagic::%>%                     automagic::make_deps_file          automagic::parse_packages          automagic::get_dependent_packages  
> automagic::parse_packages("main.Rmd")
 [1] "sdmpredictors"   "randomForest"    "summarytools"    "Boruta"         
 [5] "caret"           "raster"          "ggplot2"         "GGally"         
 [9] "rgdal"           "corrplot"        "splitstackshape" "e1071"          
> 

[nuest]

sysreqs is still not on CRAN, see r-hub/sysreqs#5