Extract strGetDummies into a dedicated module

[Copilot]

Autoloop iteration 281 intended to add strGetDummies (port of pandas.Series.str.get_dummies(sep)) as a fresh standalone function, but the symbol already lived inside the catch-all src/stats/string_ops.ts. The original auto-generated patch couldn't be applied (duplicate export, plus it used a non-existent positional new Series([...]) constructor). This PR delivers the iteration's intent: promote strGetDummies to its own module and tighten its semantics to the spec.

Implementation

• New src/stats/str_get_dummies.ts. Splits each Series element by sep (default "|"), returns a DataFrame of 0/1 indicator columns sorted lexicographically, preserves the input index.
• null / undefined / NaN (and "") now produce all-zero rows instead of falling through String(NaN) and synthesizing a stray "NaN" column.
• Columns are assembled via a Map, not a plain object, so integer-like tokens (e.g. "0") keep their lexicographic position rather than being hoisted to the front by JS object key ordering.
• All-null input still threads series.index through DataFrame.fromColumns, preserving the row count when there are zero columns.

Wiring

• Removed the duplicate strGetDummies and StrGetDummiesOptions from src/stats/string_ops.ts (and the now-unused DataFrame import / docstring entry).
• Re-export from src/stats/index.ts and src/index.ts, including the previously-missing StrGetDummiesOptions type.

Tests & docs

• Relocated the existing test block from tests/stats/string_ops.test.ts into a dedicated tests/stats/str_get_dummies.test.ts, expanded with explicit null / undefined / NaN / empty-string / index-preservation / integer-token cases, pandas-parity examples, and 5 fast-check properties (row count, 0-or-1 cells, index preservation, lexicographic columns, missing rows have row-sum 0).
• New playground/str_get_dummies.html with 4 runnable examples.

Behavioral diff

const s = new Series({ data: ["a|b", null, NaN, "b"] });
strGetDummies(s).toRecords();
// before: [{a:1,b:1,NaN:0}, {a:0,b:0,NaN:0}, {a:0,b:0,NaN:1}, {a:0,b:1,NaN:0}]
// after:  [{a:1,b:1},       {a:0,b:0},       {a:0,b:0},       {a:0,b:1}]

const s2 = new Series({ data: [" |0"] });
strGetDummies(s2).columns.values;
// before: ["0", " "]   ← Object key reorder
// after:  [" ", "0"]   ← lexicographic

[github-actions]

Commit pushed: 917c7e6

Generated by Evergreen — PR Health Keeper

[github-actions]

Evergreen fix applied ✅

The CI failure was a Biome formatter error in tests/stats/str_get_dummies.test.ts. Three multi-line expressions were formatted in a way that differed from Biome's expected output (wrapping was not applied consistently). Applied biome format --write to fix the formatting.

• Root cause: format error — Biome expected collapsed expressions on single lines
• Fix: Auto-formatted tests/stats/str_get_dummies.test.ts
• Result: Lint now passes with 0 errors (537 warnings only — all pre-existing)

Generated by Evergreen — PR Health Keeper · ● 1.3M · ◷