Remove 'exact' from diff rows wording

Author: eutwt

README.md

@@ -254,9 +254,9 @@ In full materialization, Pyversus also builds a diff table: a single
 relation with the `by` keys plus one boolean flag per value column
 indicating a difference. The table only includes rows with at least one
 difference. Those precomputed flags let row-level helpers fetch the
-exact differing rows quickly via joins, which can be faster when you
-call multiple helpers. Other modes skip the diff table and detect
-differences inline.
+differing rows quickly via joins, which can be faster when you call
+multiple helpers. Other modes skip the diff table and detect differences
+inline.
 
 - `materialize="all"`: store the summary tables and the diff table as
   temp tables. This is fastest if you will call row-level helpers

README.qmd

@@ -134,7 +134,7 @@ as DuckDB relations and are queried lazily.
 In full materialization, Pyversus also builds a diff table: a single
 relation with the `by` keys plus one boolean flag per value column
 indicating a difference. The table only includes rows with at least one
-difference. Those precomputed flags let row-level helpers fetch the exact
+difference. Those precomputed flags let row-level helpers fetch the
 differing rows quickly via joins, which can be faster when you call
 multiple helpers. Other modes skip the diff table and detect differences
 inline.

docs_api/source/getting-started.rst

@@ -51,7 +51,7 @@ in any mode; they stay as DuckDB relations and are queried lazily.
 In full materialization, Pyversus also builds a diff table: a single
 relation with the `by` keys plus one boolean flag per value column
 indicating a difference. The table only includes rows with at least one
-difference. Those precomputed flags let row-level helpers fetch the exact
+difference. Those precomputed flags let row-level helpers fetch the
 differing rows quickly via joins, which can be faster when you call
 multiple helpers. Other modes skip the diff table and detect differences
 inline.