Add complex end-to-end playground examples

[Copilot]

New Feature

What does this feature do?

Adds more complex, real-world pandas-style workflows to the Pages playground so users can explore richer end-to-end analysis scenarios in-browser.

Implementation details

• New examples

  • Added Marketplace Fraud Triage: joins transaction and merchant risk data, scores expected loss, filters review queues, and builds pivot summaries.
  • Added Inventory Replenishment Planning: combines SKU sales with product metadata, computes sell-through and reorder signals.
  • Added Marketing Attribution ROAS: reshapes spend data, joins conversions, and compares channel performance.

• Playground navigation

  • Linked all new examples from the examples gallery.
  • Updated the landing page workflow count from 10 to 13.

• Coverage

  • Added a playground conformance check to ensure the new example pages stay linked from the gallery.

Example workflow pattern:

const joined = merge(spend, conversions, {
  on: ["date", "region", "channel"],
  how: "left",
}).assign({
  roas: (df) => df.col("revenue").div(df.col("spend")),
});

[mrjf]

@copilot fix ci errors

[Copilot]

@copilot fix ci errors

Fixed in 95eb5c1. The new pandas reference snippets are now self-contained, and python scripts/validate-python-examples.py playground/ passes locally.