feat: support flagKeys param for remote evaluations

[svihang8]

Summary

This PR adds support for flagKeys param to fetch_v2() to allow emitting exposure events for designated flagKeys, in line with implementation matching core amplitude experiment sdk reference

Checklist

• Does your PR title have the correct title format?
• Does your PR have a breaking change?: No, Backwards compatible.

Hello team. Recently we ran into an issue where when setting fetch_v2(trackExposure=True), this would cause firing exposure events for all our experiments in a deployment. The workaround fix for this is to call the v1 Experiment Evaluation API directly ; reference that allows setting flag_keys explicitly to prevent firing exposure events for all keys in a deployment. From my understanding, this is supported in the node.js sdk and the core experiment sdk, so we should support it in the python sdk as well.

Cheers! 🥳

---

Note

Low Risk
Small, backward-compatible optional parameter and request header with test coverage; no auth or data-model changes.

Overview
Adds optional flagKeys on FetchOptions so fetch_v2() can ask the evaluation API to focus on specific flags (aligned with other Experiment SDKs).

When flagKeys is a non-empty list, RemoteEvaluationClient sets X-Amp-Exp-Flag-Keys to a URL-safe base64 encoding of the JSON array (padding stripped). An empty list omits the header, matching existing behavior for assignment/exposure tracking options.

Unit tests assert the header value for sample keys and that flagKeys=[] does not add the header.

^{Reviewed by Cursor Bugbot for commit 2654122. Bugbot is set up for automated code reviews on this repo. Configure here.}