Create tests root record

[mzgoddard]

Preview Tests

This change makes some changes I thought would be good to implement while working on #904.

• reduce repeated calls of path.dirname and path.basename on the same arguments
  • of note is path.basename(directory)
• base existingBuild and newBuild on the root directory instead of build directory
  • additionally update inline-scripted reference html files in the source test plan directory with FileRecordChain

[jugglinmike]

Heads up: I won't be able to review this until I can run the build script.

[jugglinmike]

This branch now conflicts with the target branch. Because this has been open for a while, there's a possibility that the conflicts are substantial, so I'd prefer to wait on reviewing until after those have been addressed.

[mzgoddard]

@jugglinmike I rebased onto master and fixed the conflicts and reproduced the changes on process-test-directory.js.

[jugglinmike]

The code looks good to me, and I've verified that its output is identical for the current "v1" and "v2" test plans.

That said, I spoke with Howard about the existing duplication in this file, and he had this to say:

v1 csv resources and references eventually won't exist so there won't be a need to build them. A couple v1 files have already been removed since introducing v2

So separating that into dedicated orchestrating scripts for the v1 and v2 test formats rather than creating branches in a single file that we may want to eventually prune in the future seemed much easier to do

Some parts are similar enough though that the functions could be consolidated in a single util and used separately

But a large majority of them felt like it would still adding a separate branch only to prune in the future from what I can remember and felt unnecessary, in that moment at least

Does that context impact that way you'd like to structure this patch?

[howard-e]

Looks good to me! Thanks for doing this work Z!

[howard-e]

... I spoke with Howard about the existing duplication in this file, and he had this to say:

v1 csv resources and references eventually won't exist so there won't be a need to build them. A couple v1 files have already been removed since introducing v2
So separating that into dedicated orchestrating scripts for the v1 and v2 test formats rather than creating branches in a single file that we may want to eventually prune in the future seemed much easier to do
Some parts are similar enough though that the functions could be consolidated in a single util and used separately
But a large majority of them felt like it would still adding a separate branch only to prune in the future from what I can remember and felt unnecessary, in that moment at least

Does that context impact that way you'd like to structure this patch?

@jugglinmike thanks for capturing my initial thoughts on this change!

Going back through this, I feel strongly now that the duplication can be significantly reduced. But it doesn't have to be done in this PR so as to unblock this. I've created #1089 to track that work.