Increase efficiency of R image build

[grallewellyn]

Takes like 2.5 hours with paws taking a long time too
cc @wildintellect

[wildintellect]

@grallewellyn can you link the current config to this ticket for easier finding.

Initial thoughts without looking:

• pick the cloud mirror (probably already did)
• rely on pre-built binaries (like rocker does, there are few choices, https://packagemanager.posit.co/)
• audit the package list (some of them can probably go upstream to py-rocket)
• make sure parallel package install is being used (Ncpu arg)
• switch to using pak https://pak.r-lib.org/

[grallewellyn]

Current R image: https://github.com/MAAP-Project/maap-workspaces/tree/develop/base_images/r

[wildintellect]

@grallewellyn I thought we were talking about the py-rocket-geospatial2 based image?

[grallewellyn]

Sorry you are right, this is the right link: https://github.com/MAAP-Project/maap-workspaces/tree/develop/base_images/2i2c/r

[wildintellect]

Ok, step one, very easy, change your mirror to https://cloud.r-project.org/