chore: raise polars floor to 1.0 and fix stale docs (#217)

The Polars engine calls LazyFrame.collect_schema(), which only exists from
polars 1.0, so every 0.20.x release the old `polars>=0.20` floor allowed failed
with AttributeError on engine="polars". Raise the floor in all five extras and
refresh uv.lock (constraints/ci.txt is unchanged).

Docs:
- data-profiling: the ydata-profiling comparison called `.plan` on a profile
  built without include_plan=True and crashed.
- trust-claims: CSV formula sanitisation is on by default on every first-party
  CSV output since 2.0.0, not opt-in.
- README quickstart output showed duplicates removed (525 -> 500); since 2.0.0
  drop_duplicates defaults to False. Match docs/quickstart.md.

Closes #213
Closes #214

Author: kevincostner17

README.md

@@ -101,12 +101,18 @@ print(report.summary())
 
 ```text
 freshdata clean report
-  rows:    525 -> 500 (-25)
+  rows:    525 -> 525
   columns: 7 -> 6 (-1)
   missing: 421 -> 0 cell(s)
-  memory:  100.8 KB -> 89.2 KB
+  actions (7):
+    - [drop_duplicates] detected 25 duplicate row(s) (4.8%), none removed
+    - [missing] 'age': filled 12 missing value(s) with median (39.6846)
+    - [outliers] 'amount': flagged 15 outlier(s) in new column 'amount_outlier'
 ```
 
+Duplicate rows are reported but kept by default; pass `drop_duplicates=True` to
+remove them.
+
 The same operation is available from the command line:
 
 ```bash

docs/data-profiling.md

@@ -86,7 +86,7 @@ sample = pd.DataFrame(
 )
 
 print("Freshdata profile preview:")
-fresh_profile = fd.profile(sample)
+fresh_profile = fd.profile(sample, include_plan=True)
 print(fresh_profile)
 print("Freshdata plan summary:")
 print(fresh_profile.plan.summary())

docs/trust-claims.md

@@ -23,7 +23,7 @@ README wording verbatim to named tests, and CI fails if either side drifts.
 | Polars/DuckDB/Spark backends scale beyond pandas | **holds with measured boundary** | see [backends](backends.md): DuckDB native handle peaks at ~40% of eager RSS (measured); `polars-lazy` defers only final materialization — pipeline intermediates still collect eagerly. Reproduce: `python benchmarks/bench_outofcore.py` |
 | Native handle kept with `strategy="conservative"` | **corrected** | also requires `fix_dtypes=False`; dtype fixing forces the recorded pandas fallback |
 | StreamingCleaner cleans arbitrarily tall data with stable memory | **holds** | bounded reservoirs/counters, RSS benchmark `benchmarks/bench_streaming.py`; dedupe window is recent-window, not global (see [limitations](limitations.md)) |
-| CSV exports safe to open in spreadsheets | **holds for review queues by default; opt-in elsewhere** | see [threat model](threat-model.md); `tests/test_csv_formula_sanitize.py` |
+| CSV exports safe to open in spreadsheets | **holds by default on all first-party CSV outputs** (`fd.clean_csv`, `freshdata clean` / `apply-plan`, the streaming CLI, review queues); opt out with `sanitize_formulas=False` / `--no-sanitize-formulas` | see [threat model](threat-model.md); `tests/test_csv_formula_sanitize.py` |
 | Typed (`py.typed`), mypy-clean | **holds** | `mypy src/freshdata` in CI quality gate (187 files) |
 | `fallback_policy="error"` prevents any unrequested pandas materialization | **holds** | enforced *before* the pandas pipeline runs at every backend delegation point; `tests/test_execution/test_fallback_policy.py` |
 | Polars subset dedup matches pandas keep semantics byte-for-byte | **holds** | order-preserving `unique(subset=...)`; parity tests incl. nulls/unicode/multi-column in `tests/test_execution/test_fallback_policy.py` (order preservation makes the stage eager, disclosed on the report) |

pyproject.toml

@@ -64,7 +64,7 @@ ml = [
 # pyarrow is required by fd.clean(polars_df): the polars→pandas interchange
 # (DataFrame.to_pandas) needs it, so the extra must ship it.
 polars = [
-    "polars>=0.20",
+    "polars>=1.0",
     "pyarrow>=10",
 ]
 pyarrow = [
@@ -84,7 +84,7 @@ bench = [
 ]
 # Everything for the out-of-core / Arrow-native execution engine.
 outofcore = [
-    "polars>=0.20",
+    "polars>=1.0",
     "duckdb>=0.10",
     "pyarrow>=10",
 ]
@@ -131,13 +131,13 @@ semantic = [
 # Everything needed for the enterprise layer (clustering, masking, lineage, CLI),
 # excluding the heavier ML label-noise extra (`cleanlab`).
 enterprise = [
-    "polars>=0.20",
+    "polars>=1.0",
     "pyarrow>=10",
     "scikit-learn>=1.0",
     "pyyaml>=5.1",
 ]
 all = [
-    "polars>=0.20",
+    "polars>=1.0",
     "pyarrow>=10",
     "scikit-learn>=1.0",
     "pyyaml>=5.1",
@@ -178,7 +178,7 @@ dev = [
     "build>=1.0",
     "twine>=5.0",
     "pre-commit>=3.0",
-    "polars>=0.20",
+    "polars>=1.0",
     "pyarrow>=10",
     "pyyaml>=5.1",
     "sqlalchemy>=1.4",