PivotTable support for multiple value fields and the "Values" axis

Cluster: pivot

Scenario

A user builds a pivot table from a tabular source and wants to aggregate several numeric measures at once, for example producing a summed value per grouping row from Initial Cost, VAT, and Total. When multiple value fields are requested and no explicit column-grouping field is given, the pivot should place the measures side by side along a synthesized "Values" axis, the way spreadsheet applications do. Today a pivot definition handles the single-measure shape but not multiple measures or the optional-columns case.

Spec note, not a corpus case: this extends an unbuilt or partial pivot feature. The durable value is the multi-measure model and the "Values" placeholder axis semantics.

Desired behavior

• A pivot definition accepts more than one value or measure field. Each supplied measure is emitted as its own data field with the chosen aggregation, such as sum.
• When two or more measures are present, the generated pivot places them along a synthesized "Values" axis, reproducing the native placeholder a spreadsheet app inserts into the column area so the measures render side by side.
• The column-grouping axis is optional: supplying no column fields is a valid pivot, and the writer defaults to the "Values" placeholder axis rather than erroring.
• The single-measure-with-column-grouping baseline, one value field distributed across a real column field's distinct values, continues to work unchanged.

Open questions

• Per-measure aggregation functions (sum, count, avg, min, max) and whether they are individually configurable, or one function for all measures initially. The aggregation-function axis, the dataField/@subtotal enum and the count-against-countNums distinction, is captured in pivot-table-aggregation-metrics.
• Where the "Values" axis sits when explicit column fields ARE present: measures nested under the column grouping, or alongside?
• Number-format inheritance: does each measure carry the source column's number format into the pivot's value cells?
• How much of this is authored against round-tripped from an application-created pivot, since the existing pivot work is passthrough-oriented (see pivot-table-parts-survive-roundtrip).

Related: pivot-table-aggregation-metrics, pivot-table-parts-survive-roundtrip, pivot-and-slicer-parts-survive-roundtrip, embedded-chart-read-write.