Introduce ubasestrip for easy consistency with SI base units

[j-fu]

The purpose is to allow users to define unitless constants like mm,nm etc
and to allow consistent calculations with them, and to provide values
of physical constants in a consistent manner.

[sostock]

Personally, I think ustrip(upreferred(q)) is not very complicated and makes it explicit what is happening, so I don’t see a reason to add a new function for it. Of course, that only works for quantities, not for units and strings. But I don’t think we should treat units like quantities anyway. But that’s only my opinion.

If we decide to add this, I would suggest to rename the function to upreferredstrip to make it explicit that it converts to upreferred (the term “base unit” might not be so clearly defined for dimensions that are not part of SI). I also would remove the String method, since no other functions except for uparse operate on strings. I think it is better to require the user to explicitly uparse the string.

[j-fu]

upreferredstrip would be ok for me. Anyway I am not intending to argue with the policy of the package.

I am working on finding a consistent pattern with Unitful for the use case in PDE codes where we sooner or later are involved with unitless packages (e.g. sparse direct solvers, mesh generators) . I would like to recommend this pattern as a policy for several packages we are developing. Unitful then would just serve
as a consistent database....

Essentially this would amount to:

const nm=upreferredstrip(u"nm")
const mA=upreferredstrip(u"mA)
const k_B=upreferredstrip(BoltzmannConstant)
...
current=20mA
... do all  calculations in numerical values of preferred units
println("current = $(current/mA) mA")

The last line here calls for something like

upreferredconvert(u::Unitful.Unitlike,v::Number)=uconvert(u,v*upreferred(u))

as the "inverse" to upreferredstrip so we could write

println("current = $(upreferredconvert(u"nA",current))")

etc

[j-fu]

I put the discussion (and another API idea) on the zulip: https://julialang.zulipchat.com/#narrow/stream/287930-Unitful Let us decide about the PR after we see where this goes...

[j-fu]

Close this for the time being - I'll try to grasp more of the unitful logic...