A `stock.move` can contain thousands of `move_line_ids` (when tracked
by SN). The union of recordset (`|=`) doesn't scale well due
to the N² complexity (for 10000 SML to validate,
). Replace recordset union
by `OrderedSet` of ids + browse (linear complexity).
Before for 10000 SML to validate:
- the `_get_in_move_lines` take 14% time in the pyflame
After in the same condition:
- the `_get_in_move_lines` isn't present anymore in the pyflame
opw-2347525
X-original-commit: 0acbf6f961e4e946c9e487343ad2cc0129400b7f
Rémy Voet (ryv)
55bf82880c