fix: make the SSE parser properly ignore non-data chunks

[hourianto]

As per https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events/Using_server-sent_events#event_stream_format and https://developer.mozilla.org/en-US/docs/Web/API/Server-sent_events/Using_server-sent_events#examples, the actual SSE stream can have so-called "comments" that do not have a field name and just have whatever comment after the colon. Any proper SSE parser should ignore those and not try to parse them.

In the case of python-genai, this is an issue when using a different base_url in http_options for a separate proxy to the Gemini API which might add extra things in the SSE stream (for example, keepalive like : this is a keep-alive). As far as I know this should be a very safe change because the official API only cares about the "data" chunks in any way (if there were any other chunks with prefixes, the json parsing would fail anyway).

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[hourianto]

Sorry for the spam in the PR, I just wanted to have a single clean commit based on the latest main