Rather than deriving the required lookup tables
from the repo metadata every time fedmod runs,
instead run the derivation code as part of the
fetch-metadata command, and cache the result as
JSON files.
There are still lots of opportunities for start-up
time optimisations here, this is mainly aimed
at separating fedmod's notion of a "working data set"
from the raw repo level metadata that DNF uses.
Rather than deriving the required lookup tables
from the repo metadata every time fedmod runs,
instead run the derivation code as part of the
fetch-metadata command, and cache the result as
JSON files.
There are still lots of opportunities for start-up
time optimisations here, this is mainly aimed
at separating fedmod's notion of a "working data set"
from the raw repo level metadata that DNF uses.