Function profileRegion

Remarks

Contentful is the one feature whose regions carry no payload: every other feature partitions on payload identity (a Fill region is one color, a Formula region one authored R1C1), so its anchor read describes the block. Occupancy has no such identity — a contentful region is pure geometry, which is why the listing could only ever print an address and a size, and why an agent had to spend one round trip per block to learn anything at all. Twelve blocks, twelve calls.

So the payload is computed rather than read: per-series counts, types and aggregates, which is df.info() + df.describe() — the shape every reader already knows, and the one openpyxl cannot produce. Nothing here is new engine work; it is getStats and getCounts per series, arranged.

Three rules keep it honest:

• A header is detected, never assumed, and the result is stated either way. A silently assumed header shifts every label by one column and the reader cannot tell.
• The longer axis is the series axis, and the profile says which. Profiling a 3×40 block by column would print forty lines describing nothing.
• Above ProfileRegionOptions.maxProfileCells the scan is a leading slice and says so — elision is explicit, per the same rule that governs every truncated listing here.