## The Problem

Validating receipts with a large number of operations by creating a backorder was timing out due to $O(N*M)$ filtering inside `_get_qty_received_without_self`, where N is the number of moves in the picking, and M is the number of moves being processed for valuation.

This happened while computing price units for moves one by one: each move filtered all picking moves to find the ones with the same `product_id`.

Another issue was a PostgreSQL `"memory exhausted"` error caused by generating a large number of OR'ed conditions, equal to the number of processed moves.

## The Solution

The massive filtering was fixed by filtering moves of the `purchase_line` instead of the `picking`, which is typically associated with only a few moves. This is still correct as the loop just after already ignores moves that don't have the same `purchase_line_id` of `self` anyways.


The PostgreSQL error was fixed by grouping moves by `location_dest_id` and generating one condition per location using an `in` clause, which is typically much smaller than generating one condition per move.

## Benchmark

Benchmark on a customer database, validating a receipt with 10k+ operations by creating a backorder:

```text
Time: timeout -> 6 min
```

OPW-6272667