[bug]: SearchClient.browse_objects only returns last page

[connesy]

Description

When using SearchClientSync.browse_objects to retrieve all records (as suggested by the documentation: https://www.algolia.com/doc/libraries/python/v4/helpers/#browse-for-records), the BrowseResponse that is returned only contains the hits from the last page. Hits on all prior pages are discarded:

client = SearchClientSync(application_id, api_key)
records = client.browse_objects(
    index_name="my_index",
    aggregator=None,
    browse_params=BrowseParamsObject(
        query="",
        attributes_to_retrieve=["some_column"],
    ),
)

print(len(records.hits))  # 227
print(len(records.page))  # 38
print(records.nb_hits)    # 38940
print(records.nb_pages)   # 39

The function that is called for every request doesn't keep the records from the previous response:
algoliasearch/search/client.py

The function _func, which is passed to create_iterable_sync, doesn't use the previous response _prev, which it gets passed in retry:
algoliasearch/http/helpers.py

The result is that only the last response from self.browse(...) is actually returned. All other responses are discarded.

I found a workaround, where I create an "aggregator" that appends the hits from each response to a non-local list, but that doesn't seem like it should be necessary:

results = []
def agg(response) -> None:
	results.extend(response.hits)

client.browse_objects(
    index_name="my_index",
    aggregator=agg,
    browse_params=BrowseParamsObject(
        query="",
        attributes_to_retrieve=["some_column"],
    ),
)

print(len(results))  # 38940

### Client

Search

### Version

4.6.2

### Relevant log output

_No response_

[shortcuts]

Hey, thanks for opening the issue! Actually we should make it clear that aggregator is required, the workaround you found is in fact the way to leverage this browse_objects helper, we do the pagination and you use the responses however you want