Refactor manifest creation in New-FinOpsTestData to reduce duplication

Summary

Refactor the 4 manifest creation sites in New-FinOpsTestData to use a shared helper, reducing code duplication and schema drift risk.

Motivation

New-FinOpsTestData currently creates manifest JSON in 4 separate locations — one per dataset type that has a slightly different schema (Costs, Prices, CommitmentDiscountUsage, Recommendations/Transactions). Each site constructs the manifest hashtable independently, which means:

• Watermark fields (_ftkTestData, _generator, _generatedAt) must be added in 4 places
• Schema changes require updates in 4 places
• Risk of drift between manifest formats

Proposed changes

• Extract a private helper function (e.g., New-TestDataManifest) that accepts dataset-specific parameters and returns the manifest hashtable
• Each call site passes only the dataset-specific values (dataset name, blobs, row count, export config overrides)
• The helper handles common fields: blobCount, byteCount, dataFormat, runInfo, _ftkTestData watermarks
• This is a prerequisite/companion to #2016 (New-FinOpsExportManifest)

Context

Deferred from PR #2006 (New-FinOpsTestData and Remove-FinOpsTestData). Can be combined with #2016 if extracting to a public command.

Start at the New-FinOpsTestData entry point and locate its four manifest creation sites for Costs, Prices, CommitmentDiscountUsage, and Recommendations/Transactions. Compare the shared fields and dataset-specific parameters, then review #2016 and the context from PR #2006. Done means one private helper supplies the common manifest fields while all four call sites preserve their dataset-specific schemas.