Implement `LinearAlgebra.norm` where appropriate

[tpgillam]

e.g. for GA30 it would be valid to say:

LinearAlgebra.norm(v::GA30.Odd) = sqrt(dot(v, v))

But I think this holds more generally for all algebras(?). Currently attempting to compute a norm fails, since e.g. GA30.Odd <: Number, so we fall back to an implementation that attempts to cast the multivector to a float, which naturally fails.

[tpgillam]

Note that doing this should give us LinearAlgebra.normalize for free, since it is implemented in terms of norm.