Move "raw" bundles to Data.X.Base

[MatthewDaggitt]

Currently the raw bundles, e.g. RawMagma are defined in Data.X.Properties but they should really be defined in Data.X.Base as they don't require any properties.

[JacquesCarette]

Except that I see that this makes Data.X.Base now depend on Algebra.Bundles. That's not too too heavyweight, but do we want Data.X.Base to have quite so much in it? I'd prefer Data.X.Bundles, now that I look at the details of #1810 .

[Taneb]

I'd rather something like a new module Algebra.RawBundles or Algebra.Bundles.Raw containing the definitions of the raw bundles, which can be imported at much lower cost from Data.X.Base

[JacquesCarette]

But that would still leave Data.X.Base to depend on Algebra.Bundles if those are actually imported.

Won't Algebra.RawBundles (or equivalent) grow to something really large, as it would depend on all algebras? I certainly think that having these be defined in a module that mentions X in its name would help.

[Taneb]

I think I didn't explain myself right. I meant that the definition of the types RawMagma, RawMonoid, RawGroup, etc, be moved into the new module Algebra.RawBundles, and then Data.X.Base can import Algebra.RawBundles and we define X-rawGroup in Data.X.Base.

[MatthewDaggitt]

I agree with @Taneb (and would vote for Algebra.Bundles.Raw)

[Taneb]

I'm going to try to make a PR this evening moving the definitions of RawX to Algebra.Bundles.Raw

[5xy1]

@Taneb Maybe I could help with this? I can update my PR today.