llms/googleai: batch embedding calls

[johanbrandhorst]

The new Batch API significantly speeds up embedding

PR Checklist

• Read the Contributing documentation.
• Read the Code of conduct documentation.
• Name your Pull Request title clearly, concisely, and prefixed with the name of the primarily affected package you changed according to Good commit messages (such as memory: add interfaces for X, Y or util: add whizzbang helpers).
• Check that there isn't already a PR that solves the problem the same way to avoid creating a duplicate.
• Provide a description in this PR that addresses what the PR is solving, or reference the issue that it solves (e.g. Fixes #123).
• Describes the source of new concepts.
• References existing implementations as appropriate.
• Contains test coverage for new functions.
• Passes all golangci-lint checks.

[tmc]

So good to see you contributing! can't wait to get this in!.

[johanbrandhorst]

Long time no see Travis :). Well done on this project. I'm wondering how you would prefer I test this, the only way I can see is to spin up a custom HTTP server to fake the requests but it doesn't provide much in the way of confidence - I have tested this personally against the Gemini API and it works great.

[tmc]

Certainly open to ideas on improving the testing strategy for this, putting this in now but do consider a test as you see fit.

[eliben]

Also happy to see you contribute here, @johanbrandhorst :-)

Re tests, the googleai provider is relatively well-tested with live tests vs. the Gemini API (https://github.com/tmc/langchaingo/blob/main/llms/googleai/shared_test/shared_test.go)

For this test specifically, testing batching vs. no batching is a bit tricky and we don't really do mock testing with fake HTTP backends for now. That said, if the shared live tests pass and you observe the performance improvement, I believe this is good enough.