Fix incorrect results in bruteforce with filter

[lukaszsmolinski]

There is a bug in the searchKnn function of BruteforceSearch. When it is used with a filter, in some cases, it returns fewer items than expected. This PR should fix it.

The bug can be reproduced using the following code:

import hnswlib

data = [[0], [1], [2]]
max_elements = len(data)
dim = len(data[0])

bf_index = hnswlib.BFIndex(space='l2', dim=dim)
bf_index.init_index(max_elements=max_elements)
bf_index.add_items(data, ids=range(max_elements))

filter_function = lambda id: id in [0, 2]
labels, _ = bf_index.knn_query(data[0], k=2, filter=filter_function)

print(labels)
# Should be: [[0 2]]
# Is: [[0 0]]

[dyashuni]

@lukaszsmolinski Thank you for the PR!
Good catch. Looks good.

[dyashuni]

For some reason, this code

std::priority_queue<std::pair<dist_t, hnswlib::labeltype >> result = alg->searchKnn(
    (void*)items.data(row), k, p_idFilter);
for (int i = k - 1; i >= 0; i--) {
    auto& result_tuple = result.top();
    data_numpy_d[row * k + i] = result_tuple.first;
    data_numpy_l[row * k + i] = result_tuple.second;
    result.pop();
}

from here works even if result.size() < k.

Could you please add an exception if result.size() != k to prevent unexpected results from being returned?
Something like this

if (result.size() != k)
  throw std::runtime_error(
      "Cannot return the results in a contiguous 2D array. Probably ef or M is too small");

from here

[lukaszsmolinski]

Done. By the way, it seems like this function is also missing the if (normalize) check. I can fix that in this PR if you'd like.

[dyashuni]

Thank you, @lukaszsmolinski !
Yes, please add query normalization to BFIndex search to align it with Index

[dyashuni]

Thank you!