fix(cdp): base64-encode non-UTF8 Network.getResponseBody payloads

[gilangjavier]

Summary

• return plain text with base64Encoded=false when the captured response body is valid UTF-8
• base64-encode non-UTF8 payloads and return them with base64Encoded=true
• add regression tests covering both UTF-8 and binary response paths

Fixes #1903.

[github-actions]

All contributors have signed the CLA ✍️ ✅
_{Posted by the CLA Assistant Lite bot.}

[gilangjavier]

I have read the CLA Document and I hereby sign the CLA

[krichprollsch]

Hello @gilangjavier,
Thanks for the contribution.

I'm wondering if we could do that easier: always return a base64 encoded value, even for utf-8 responses.
We could then directly encode into base64 when we allocate to save the captured response and save an additional allocation on getResponseBody.

If we really want to keep utf-8 value non encoded, I guess we have to change captured_responses to store a struct containing a base64encode bool and the data. We can add the the entry and set base64encode field by analyzing the resp mime type + an .empty data during the header done callaback.

And adapt the the captured response callback to save the data depending base64encode.

[gilangjavier]

Thanks for the detailed suggestion. I updated the PR to use the simpler approach: Network.getResponseBody now always returns base64-encoded payloads with base64Encoded=true, including UTF-8 responses.\n\nI also updated the tests accordingly:\n- UTF-8 response now asserts aGVsbG8= with base64Encoded=true\n- binary response test remains covered\n\nPushed commit b5b012b to this branch. Please re-check when you have time.