Consider renaming all the `_code_system` aliases to just `_system`

[mikix]

When we did the big field rename for 2.0 that aligned us closer to the FHIR spec, we used the following scheme for codeable fields:

    interpretation_code,
    interpretation_code_system,
    interpretation_display,

But I personally find that a little confusing in a world where we are mostly mapping the FHIR spec. Why not use just system, which is what the field is called. code_system makes my brain think we're talking about code.system instead of system.

So my pitch is that for 3.0, we consider a mass rename back like s/code_system/system/g as one of its breaking changes.

[dogversioning]

For postertity, documenting our off-thread discussion:

• We are going to continue using code_system for new tables in 2.0, and then make the change in 3.0
• For studies we control, we should consistently use the pattern baseResource_code_system when joining into tables so it will be easier to bulk replace going forward.