feat(csv): column statistics popover (counts, distinct, numeric summary, top values)

[salmonumbrella]

Problem

Inspecting a CSV column today means scrolling the data, eyeballing values, and maybe running Copy Column Values into a shell to count things. A native column-level profile would close the biggest "why bounce to Modern CSV" gap — their Column Analysis and Unique Values popovers are the single most-cited reason CSV users keep that app around alongside a database client.

#1454's per-column local filter already needs distinct-value enumeration internally; the same primitive can drive a profile popover.

Proposed solution

Right-click a column header → Column Statistics… opens a popover showing, for the currently loaded rows:

• Counts — total, non-empty, empty
• Distinct value count
• Detected type — string, int, float, date, mixed
• Numeric summary — min, max, mean, median, sum, when the type is numeric
• Top-N frequent values — value, count, percentage, inline frequency bar

The popover should:

• Open from both the column-header context menu and an Edit or View menu entry
• Stay open while the user scrolls so values stay cross-referenceable
• Re-compute when filters from feat(datagrid): per-column local value filter (Excel-style) #1454 change the visible row set

Implementation notes

• The DuckDB plugin can back this with SELECT … FROM read_csv_auto(…) GROUP BY column queries when available; pure-Swift counters work for the in-memory path.
• feat(datagrid): per-column local value filter (Excel-style) #1454's distinct-value list is the same primitive at a different surface; consider sharing the implementation.