Add option to repeat queries in ground truth

[mihaic]

No description provided.

[Copilot]

Pull Request Overview

This PR introduces an option to repeat query vectors when generating ground truth, updating both the CLI and associated test suites to support this new functionality.

• Added new tests in test_generate_ground_truth.py for scenarios with shuffling and repeated queries.
• Updated conftest.py with new constants and modified query file generation.
• Enhanced generate_ground_truth in src/svsbench/generate_ground_truth.py to handle repeated queries via the new num_query_vectors parameter.

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
tests/test_generate_ground_truth.py	New tests covering no-shuffle, shuffle, and repeated query scenarios.
tests/conftest.py	Added constants and updated query path generation.
src/svsbench/generate_ground_truth.py	Added new parameters and logic to optionally repeat queries.

[Copilot]

Consider adding inline comments within the while-loop block that repeats queries to clarify its logic and purpose, which can help future readers understand the batch-based repetition process.

Suggested change

	cursor = 0
	while cursor < num_query_vectors:
	permutation = rng.permutation(len(queries))
	batch_size = min(num_query_vectors - cursor, len(queries))
	queries_all[cursor : cursor + batch_size] = queries[
	permutation[:batch_size]
	]
	ground_truth_all[cursor : cursor + batch_size] = idxs[
	permutation[:batch_size]
	]
	cursor = 0
	# Repeat the process until we have generated the required number of query vectors.
	while cursor < num_query_vectors:
	# Generate a random permutation of the query indices to shuffle the queries.
	permutation = rng.permutation(len(queries))
	# Determine the size of the current batch, ensuring we don't exceed the total required.
	batch_size = min(num_query_vectors - cursor, len(queries))
	# Select a batch of queries based on the permutation and add them to the output array.
	queries_all[cursor : cursor + batch_size] = queries[
	permutation[:batch_size]
	]
	# Select the corresponding ground truth indices for the batch and add them to the output array.
	ground_truth_all[cursor : cursor + batch_size] = idxs[
	permutation[:batch_size]
	]
	# Update the cursor to reflect the number of queries processed so far.