base64 dense vector packing

[miguelgrinberg]

Elastic\Elasticsearch\Helper\Vectors::packDenseVector() function to pack dense vectors for efficient bulk uploading to Elasticsearch. Also included in this PR is a benchmark for this feature.

[picandocodigo]

I think we're supposed to leave out the docid in the benchmark.

[miguelgrinberg]

I believe that refers to not using the docid as the document id. This puts it as a regular field, which I saw others do as well. I don't think it will make any significant difference with or without this field, anyway.

[pquentin]

For small dataset sizes, it likely doesn't matter, but setting a doc _id can lead to reduced performance because for each insert, Elasticsearch needs to check if there's already a doc with that _id.

[miguelgrinberg]

@pquentin this is a docid field added in the document, not the metadata _id.

[ezimuel]

LGTM, thanks for the PR!

[github-actions]

The backport to 9.3 failed:

The process '/usr/bin/git' failed with exit code 128

To backport manually, run these commands in your terminal:

# Fetch latest updates from GitHub
git fetch
# Create a new working tree
git worktree add .worktrees/backport-9.3 9.3
# Navigate to the new working tree
cd .worktrees/backport-9.3
# Create a new branch
git switch --create backport-1499-to-9.3
# Cherry-pick the merged commit of this pull request and resolve the conflicts
git cherry-pick -x --mainline 1 d06ae765a78d0d4a2a6e3ed6dac68e219cc614c1
# Push it to GitHub
git push --set-upstream origin backport-1499-to-9.3
# Go back to the original working tree
cd ../..
# Delete the working tree
git worktree remove .worktrees/backport-9.3

Then, create a pull request where the base branch is 9.3 and the compare/head branch is backport-1499-to-9.3.