Currently, when an operation is loaded in the barcode app, the frontend executes a request to get_barcode_data which will perform a read on all relevant barcode data from the backend to cache it on the frontend. The problem is, when loading the records of `stock.quant.packages`, it will attempt to read all empty packages with no location id. There are cases where these packages are way too many i.e. 500K and thus to read fields on these records, it would be terribly slow. As this is done upon opening every operation in the barcode app, it's making the app feel sluggish.

To workaround the problem, this PR introduces a config paramter `stock_barcode.usable_packages_limit` which will limit the total number of empty packages being read to feed the frontend cache. This gives the option to tradeoff between cache locality and network connection. Setting this limit to a smaller number will result in more rpc calls instead of suffering a slowdown on one big rpc call to load all data once.

Benchmarks:

This will totally depend on the configured limit. Lower limit would mean higher speedup and vice versa. As an example, setting a limit of 1000 cut down the loading of get_barcode_data from 13.87 s to 768 ms

opw-4245980

Forward-Port-Of: odoo/enterprise#74096

### Steps to reproduce the issue:

1. Create an Invoice with "ODOO_BUG" as payment reference
2. Create a Bank Transaction with the same amount and customer as the Invoice and "ODOO_BUG" as Label
3. The Transaction and the Invoice are automatically matched
4. Create an Invoice with "ODOO BUG" as payment reference
5. Create a Bank Transaction with the same amount and customer as the Invoice and "ODOO BUG" as label
6. The Transaction and the Invoice are not automatically matched

### Explanation:

When evaluating `st_line_text_values` in `_get_invoice_matching_st_line_tokens`, the values are split at every whitespace. When trying to find `account.move.line` that match the tokens resulting from the operations, every `name`, `move_id.name` or `move_id.ref` with a whitespace in it will not be recognized due to the whitespaces being removed from the `account.bank.statement.line` side.

### Fix reasoning:

From a business perspective, no Invoice/Payment/Bill reference should have a space in it bar some already covered exceptions. To avoid performance heavy queries, the text value will simply be added to `exact_tokens` and treated as such, allowing for an automatical reconciliation between an `account.move.line` and an `account.bank.statement.line` if their refs match perfectly.

opw-4114296