Demo: annotating hand edits to an importer-generated CSV namespace

[SimonCockx]

Illustration only — not intended to merge. Raised to show what the annotate, don't block guard looks like on a real pull request, for the STORY-1911 CSV import design.

Setup (on the base branch, not in this diff)

demo.unavista.csv is an example namespace of the kind model-import would generate from CSV samples. Its rune-config.yml entry carries an origin marker and is deliberately not read-only:

namespaceConfig:
- namespace: demo.unavista.csv
  origin:
    modelImport: csv

A new workflow, annotate-generated-namespaces.yml, reports on pull requests that touch such namespaces. It never fails — refining a generated model by hand is the intended workflow, so a blocking check would put a bypass label on nearly every PR and the signal would decay into a ritual.

What this PR changes

Two ordinary refinements a modeller would make after an import:

File	Edit	Level
unavista-csv-type.rosetta	renames the [label] on tradingDateTime to match the source column, and documents it	warning
unavista-csv-enum.rosetta	expands the documentation on MTCH	notice

The label change is raised to warning because the CSV column binding depends on labels — that is the class of edit that can silently break ingestion. Documentation changes stay at notice level, so the warning keeps its meaning.

What to look at

• Files changed — annotations appear inline against the edited lines.
• The workflow run summary — a table of every generated namespace the PR touches.
• The check itself is green. Visibility, not enforcement; correctness is left to review and to an ingestion test pack.

Note: the model omits a schema <name> CSV declaration, since this repo is on DSL 10.2.0 which predates that construct.